THE
RABBIT

HOLE

The indefinitive, perpetually outdated, playbook
for new and emerging product leaders.

**Marty Ringlein** *&* **Will Hubbard                                                                                                                                        mpd²**

# Predictable Product Development

Designing outsized returns through early-stage experimentation and power law principles.

Introduction: From an intro to Product Development to the future of Product Development. This document explores the foundations for how transformative products are conceived, validated, and scaled in environments of uncertainty. The following integrates the history and evolution of product management with modern experimentation practices, data-driven decision making, and product-led growth strategies.

Every product roadmap is a collection of bets. We’re betting that a problem matters, that our solution will solve it, and that someone will care enough to change their behavior or pay for it. Those bets consume time, money, and attention long before we know whether we’re right. How we place them, test them, and decide what deserves more investment determines how much opportunity we give ourselves to build something meaningful.

That’s the idea behind **Predictable Product Development**. We can create a repeatable way to explore opportunities, challenge assumptions, and improve our odds of success. Any individual idea can fail. The opportunity is to build a system that keeps learning and gives the most promising ideas room to become something much larger. So large in fact, that the best ideas have such an outsized return on investment that they more than pay for the ideas that are less successful.

Venture capital offers a useful way to think about this. Investors operating under power law principles expect a small number of investments to generate a disproportionate share of their returns. They build a portfolio, accept that many bets won’t work, and reserve capital to invest further when the evidence warrants it. One exceptional outcome can matter more than dozens of modest successes.

Product teams can apply that same thinking. Give yourself more shots on goal. Explore different approaches to a meaningful problem. Keep the initial investment small enough that being wrong is affordable. Know what you need to learn and what would justify the next round of investment. When an opportunity starts showing real promise, have the resources and conviction to pursue it.

This requires discipline. Ten versions of the same untested assumption won’t teach you much. Neither will a hundred experiments whose results never change a decision. Each test should answer a question that matters, and what you learn should influence what happens next. Sometimes that means refining the idea. Sometimes it means stopping. Sometimes it means recognizing that an unexpected result is more interesting than the thing you set out to prove.
AI changes the economics of this work. Teams can explore more alternatives, create higher quality prototypes, and put working experiences in front of people with considerably less time and expense. An idea that once required weeks of coordination can often become something testable in hours. Shorter cycles give us more opportunities to learn within the same budget, while more realistic prototypes let us test behavior that a slide or sketch could only invite people to imagine.

The principles and frameworks in this playbook help make those decisions. First principles thinking challenges the assumptions we inherited. Customer research and behavioral insights help us understand what people actually need and why they act the way they do. Experimentation and feedback loops reveal where our expectations hold up. Prioritization helps us decide where to invest next. Positioning, distribution, and partnerships help turn an early success into a business that can grow.

Use these tools where they help you move forward, faster. Question them when they don’t. Keep the products people already depend on working well, while making room to explore what might come next. Be willing to pursue an idea before it becomes obvious, and equally willing to change your mind when the evidence changes.

**Predictable Product Development** is a practice of earning conviction. Start with a clear problem, make a small bet, learn something useful, and let that learning shape the next decision. Over time, you get better at choosing your shots and recognizing which ones deserve everything you’ve got.

# We’ve built products from zero-to-one, through IPO, and within the Fortune 100

This playbook is a culmination of over two decades of experience as successful entrepreneurs, corporate executives, venture capitalists, and professors.

Hello! Marty and Will have been working together for over a decade. They’ve started and sold multiple companies, raised tens-of-millions in venture capital, and have invested early in some of the most successful start-ups in history — including Figma, SpaceX, and Anthropic. They also led innovation initiatives at some of the largest companies in the world — including Twitter, Apple, Disney, Oracle, and The White House.  In addition to being Adjunct Professors at Northwestern University, Marty and Will are General Partners in the Adventure Fund where they invest together in early-stage startups as well as co-founders in Agree.com, a Silicon Valley based seed-stage startup. They also apply their experience and this playbook to help executive leadership teams with R&D and corporate innovation initiatives, such as Kraft Heinz, P&G, and Northwestern Mutual.

|  | Marty Ringlein Marty studied Computer Science at the University of Maryland and Business at Columbia Business School. He quickly fell in love with design, spending much of his early career working as both a designer and developer. After his first company was acquired by Twitter, he went on to sell his second to Eventbrite and his most recent to Brex. And he’s still going. [https://linkedin.com/in/martymadrid](https://linkedin.com/in/martymadrid)  |
| :---- | :---- |
|  |  |
|  | **Will Hubbard** Will designed his own major at UC Berkeley in "Entrepreneurship, Capital Investment & Economic Development" and has been living it ever since. He co-founded his first company before graduating, sold it, and then built and sold two more. Today he's building Agree.com and investing in founders as a General Partner at Adventure Fund. [https://linkedin.com/in/wbhub/](https://linkedin.com/in/wbhub/)  |

# *Almost* Everything you need to know

*The rabbit-hole went straight on like a tunnel for some way, and then dipped suddenly down, so suddenly that Alice had not a moment to think about stopping herself before she found herself falling down a very deep well.  - Alice's Adventures in Wonderland*

| 01 | [Introduction [Foreword]](#predictable-product-development) | 64 | [Dimensionalities of Data](#dimensionalities-of-data) |
| :---: | :---- | :---: | :---- |
| **05** | [The History of Product Management](#the-history-&-evolution-of-modern-product-management) | **67** | [Data Co-opetition](#data-co-opetition) |
| **07** | [First Principles Thinking](#first-principles-thinking) | **70** | [Metcalfe's Law](#metcalfe's-law) |
| **10** | [Human-Centered Design](#human-centered-design-principles) | **73** | [Human-in-the-Loop](#human-in-the-loop) |
| **13** | [Agile Development Fundamentals](#agile-development-fundamentals) | **76** | [Gen AI & Agentic Workflows](#generative-ai-and-agentic-workflows) |
| **16** | [User Personas & Segmentation](#user-personas-&-segmentation) | **79**  | [Product Positioning [The 4 P’s]](#product-positioning-[the-4-p’s]) |
| **19** | [Jobs-to-be-Done [Framework]](#jobs-to-be-done-[framework]) | **82**  | [SWOT Analysis [Framework]](#swot-analysis-[framework]) |
| **22**  | [Innovation Adoption Curve](#innovation-adoption-curve) | **85** | [Backcasting [Framework]](#backcasting-[framework]) |
| **25** | [Behavioral Economics](#behavioral-economics) | **88** | [Kano Model [Framework]](#kano-model-[framework]) |
| **28** | [Cognitive Biases](#cognitive-biases) | **91** | [R.I.C.E. [Framework]](#r.i.c.e.-[framework]) |
| **31** | [Maslow’s Hierarchy of Needs](#maslow's-hierarchy-of-needs) | **94** | [Strategic Sourcing [Framework]](#strategic-sourcing-[framework]) |
| **34** | [Prediction Mechanisms](#prediction-mechanisms) | **97** | [MSCW Method](#mscw-method-[framework]) |
| **37** | [Power Law Distribution](#power-law-distribution) | **100** | [PESTLE [Framework]](#pestle-[framework]) |
| **40** | [J Curve](#j-curve-[portfolio-theory]) | **103** | [RFM [Framework](#rfm-[framework])] |
| **43** | [Barbell Theory [Blue Ocean Strategy]](#barbell-theory-[blue-ocean-strategy]) | **106** | [RACI Matrix](#raci-matrix) |
| **46** | [Pareto Principle](#pareto-principle-[80/20-rule]) | **109** | [New & Emerging Technologies](#new’ish-&-emerging-technologies) |
| **49** | [Data-Driven Decision Making](#data-driven-decision-making) | **113** | [Tools of the Trade](#tools-of-the-trade) |
| **52** | [Continuous Delivery Fundamentals](#continuous-delivery-fundamentals) | **116** | [The Anatomy of a PRD](#the-anatomy-of-a-prd) |
| **55** | [Real-Time Feedback Loops](#real-time-feedback-loops) | **119** | [Acronym Cheat Sheet](#acronym-cheat-sheet-(a.c.s.)) |
| **58** | [The Data Feedback Loop](#the-data-feedback-loop) | **123** | [Book Club [Recommended Reading]](#book-club-[recommended-reading]) |
| **61** | [Data Radar [Framework]](#data-radar-[framework]) | ∞  | ***… never stop learning.*** |

# The History & Evolution of Modern Product Management

Introduction: Product management wasn't invented in Silicon Valley — it was assembled, piece by piece, over about ninety years. Brand management gave it customer ownership, Toyota gave it continuous improvement, IDEO gave it problem framing, and Agile gave it permission to change its mind. Here's the lineage, and what each era left behind in the job you do today.

|  | 1931 — The P&G Brand Memo The now famous three-page memo written by Neil H. McElroy at P&G became a cornerstone in modern thinking about brand management and ultimately product management; this put decision making as close as possible to the customer, making the product manager the voice of the customer. |
| :---: | :---- |
|  | 1943 **— The P&G way meets the Hewlett-Packard Way** McElroy advised at Stanford where he influenced two young entrepreneurs, Bill Hewlett and David Packard. At HP, there was a fundamental evolution from one of brand management to one of product management. They also introduced the division structure, where each product group became a self-sustaining organization. |
|  | 1948 **— Toyota Production System Developed** In post-war Japan, industries develop just-in-time manufacturing. Eiji Toyoda brought two fundamental ideas to the forefront: Kaizen – improving the business continuously while always driving for innovation and evolution and Genchi Genbutsu – to go to the source to find the facts to make correct decisions. |
|  | 1970 **— Waterfall Method Popularized** The Waterfall process was heavy-handed and process-oriented, often resulting in product development being a slow and laborious process, even in the tech industry. |
|  | 1991 **— IDEO introduces ‘Design Thinking’** Rather than accept a product problem as given, product managers now explore the given problem and its context and may re-interpret or restructure the given problem in order to reach a particular framing of the problem that suggests a route to a solution. |
|  | 2001 **— Agile Manifesto Authored** The Agile Manifesto became a watershed moment in product development, freeing up Product Management from focusing on deliverables like specs to focusing on customer collaboration. This focus shift was profound on many levels; changing the relationship between Product and Engineering from an adversarial one to a collaborative one. |
|  | 2011 **— Eric Ries publishes ‘The Lean Startup’** Ries unknowingly provided a blueprint for corporate innovation that shifted the focus from just building products to building products that people genuinely want and need, underlining the importance of adaptability, rapid feedback, and continuous learning in the unpredictable world of startups and product development. |
|  | 2023 **— Predictable Product Development** Traditional product development was geared towards getting products to market faster but is now shifting significantly to R&D-based development. The formal product development process outshines the unpredictable nature of informal, hit-or-miss processes. Focused attention to process and detail, with adaptive openness to possibilities, is increasingly necessary for success. |

**Continue Down the Rabbit Hole …**

* **Evolution of the product manager** [*ACM Digital Library*]
  [https://dl.acm.org/doi/10.1145/2669480/](https://dl.acm.org/doi/10.1145/2669480/)

* **The history and evolution of product management** [*Mind the Product*] [https://mindtheproduct.com/history-evolution-product-management/](https://mindtheproduct.com/history-evolution-product-management/)

* **Product managers for the digital world** [*McKinsey*] [https://mckinsey.com/industries/technology-media-and-telecommunications/our-insights/product-managers-for-the-digital-world/](https://mckinsey.com/industries/technology-media-and-telecommunications/our-insights/product-managers-for-the-digital-world/)

# First Principles Thinking

Introduction: The way a product works today is the result of decisions someone made, often under constraints that may no longer apply. First principles thinking means breaking a problem down to its underlying facts, separating those facts from assumptions, and building a solution from that foundation. In Predictable Product Development, it helps us question what actually needs to happen for a customer to get the outcome they want. This gives us room to reconsider the product itself, including its features, costs, and workflow, before committing to another version of what already exists.

**The Need to Know**

Product teams inherit assumptions all the time. A purchase requires an account. A process needs three approvals. A device needs a screen. Sometimes there are good reasons for these decisions. Sometimes the explanation is simply that the original product worked that way. When those assumptions go unchallenged, teams can spend months improving a process that could have been simplified or removed.

First principles start by defining the outcome independently of the existing solution. Imagine a team asked to make a ten-field registration form easier to complete. Before rearranging the fields, they could determine what information is actually required, why it is needed, and when it must be collected. Perhaps only two fields are necessary to get started. That possibility becomes visible once the team examines the requirements behind the form.

The discipline is knowing the difference between a fact, a constraint, and an assumption. Physical limits, demonstrated customer needs, and verified business requirements deserve different treatment from habits or untested beliefs. “Customers won't trust this” is a hypothesis to investigate. “Our current system can't do this” describes a technical constraint whose cost and flexibility need to be understood. Neither statement should automatically end the conversation.

**The Real-world Application**

**Dyson’s** bagless vacuum was invented out of first principle thinking. James Dyson investigated why his vacuum lost suction and found that dust clogged the bag. He had also encountered industrial cyclones that separated dust from air without a membrane filter. Applying that separation method to a household vacuum opened another approach to the problem. In an interview, Dyson described investigating an industrial cyclone and building one for his factory before applying the idea to vacuum cleaners.

Viewed through first principles, the essential requirement was to move dirt off the floor, separate it from the airflow, and collect it. A disposable bag was one way to accomplish that. Once the bag became a design choice that could be questioned, the team could explore a different mechanism. This also shows that first principles thinking can draw on existing technology. The originality came from reconsidering which solution belonged in the product.

**How this Applies to You**

As a product leader, it is our responsibility to be able to explain why a requirement exists. When someone asks for a feature, investigate the problem behind it and the conditions a successful solution must satisfy. That responsibility calls for enough customer, technical, and business understanding to challenge a proposal thoughtfully. A useful question is: “What would happen if we removed this, and what evidence tells us that?”

You also need to make your reasoning available for others to challenge. State which facts you know, which assumptions you are making, and what would change your mind. Bring engineers, customers, and relevant business partners into that work early. A solution that looks obvious on a whiteboard may depend on an assumption that someone else can disprove in five minutes.

Use this approach where an inherited assumption materially limits the customer experience, cost, or performance. Familiar conventions can still be the right choice, particularly when they help people understand and use a product. Your job is to know why you are keeping them. Before improving the next screen, feature, or process, ask: What actually needs to happen here, and why have we decided it has to happen this way?

**Continue Down the Rabbit Hole …**

* **What is First Principles Thinking?** One of the best ways to discover new solutions. [*Farnam Street*]
  [https://fs.blog/first-principles/](https://fs.blog/first-principles/)

* **Design Thinking Bootleg:** An introductory set of tools and methods for approaching a new project. [Stanford *d.School*]
  [https://dschool.stanford.edu/tools/design-thinking-bootleg/](https://dschool.stanford.edu/tools/design-thinking-bootleg/)

* **Subtract:** Why getting to less can mean thinking more [*Behavioral Scientist*]
  [https://behavioralscientist.org/subtract-why-getting-to-less-can-mean-thinking-more/](https://behavioralscientist.org/subtract-why-getting-to-less-can-mean-thinking-more/)

* **Think Like a Rocket Scientist:** The habits, ideas, and strategies that turn the seemingly impossible into the possible. [*Ozan Varol*]
  [https://ozanvarol.com/rocket-scientist](https://ozanvarol.com/rocket-scientist)

# Human-Centered Design Principles

Introduction: A product can work exactly as intended and still fail the person using it. Human-centered design starts with understanding people, their circumstances, and what they are trying to accomplish. It brings them into the design process through observation, conversations, prototypes, and testing. In Predictable Product Development, this helps us discover whether we are solving a meaningful problem and whether our solution works in someone’s actual life. The goal is to make understanding the customer a continuous input into product decisions, from defining the problem through improving the experience after launch.

**The Need to Know**

Product teams know too much about their own products. We understand the terminology, remember where features live, and know which steps to follow. Customers arrive with different expectations, abilities, and levels of experience. Something obvious to the team may be confusing to someone encountering it for the first time. Human-centered design helps expose that gap by learning directly from the people who will use the product.

The core principles are practical: understand people in context, involve them throughout development, consider different abilities and circumstances, evaluate the whole experience, and iterate based on evidence. Observation helps reveal problems people may have stopped noticing or struggle to explain. Prototypes give them something concrete to try. Testing gives the team a reason to revise its assumptions. IDEO.org describes this as a process of learning from people, developing and sharing possible solutions, and bringing those solutions into use.

This also changes how you interpret a feature request. Someone asking for a spreadsheet export may need to share results with a colleague who cannot access the product. An export might be the right answer, but understanding the situation opens other possibilities. You can investigate permissions, collaboration, or scheduled reporting before committing to a solution. The customer provides essential knowledge about the problem; the team remains responsible for developing and testing a workable response.

**The Real-world Application**

The initial idea behind **OXO’s** Good Grips kitchen tools grew from Betsey Farber’s difficulty using a conventional vegetable peeler because of arthritis. Her husband, Sam Farber, saw an opportunity to improve the tool. Introduced in 1990, Good Grips used thick, nonslip handles with flexible fins to make kitchen utensils easier to hold. The design addressed a difficulty that conventional products had left unresolved.

The useful design lesson is how a specific person’s difficulty can reveal a broader opportunity. A handle that requires less effort to grip can also help someone with wet hands or someone preparing food for an extended period. Viewed through human-centered design, the handle’s shape and material become central to the product’s usefulness. Those benefits still need testing with different users and conditions. One person’s experience can expose a problem without representing everyone who will use the solution.

**How this Applies to You**

It is imperative to treat our understanding of the customer as something that needs evidence. Empathy is a starting point, but imagining how someone feels can still leave you designing around your own assumptions. Make room for direct observation and let customers show you where your explanation of their behavior is incomplete.

Your responsibility also extends across the experience. A clear interface cannot compensate for confusing pricing, difficult installation, or an unresolved support issue. Work with engineering, operations, sales, and support to understand how their decisions affect the person using the product. Business constraints matter, and part of your job is finding a solution that serves people while remaining feasible to build and sustain.

Before calling a design successful, ask: Who can use this successfully, under what conditions, and who still struggles? Measure whether people complete the task, how much effort it takes, and whether they can recover from mistakes. A polished demonstration is useful. Seeing someone accomplish what they came to do is stronger evidence that the design works.

**Continue Down the Rabbit Hole …**

* **In the Age of AI, Human-Centered Design Is Crucial** [*IDEO*]
  [https://ideo.com/journal/in-the-age-of-ai-human-centered-design-is-more-crucial-than-ever](https://ideo.com/journal/in-the-age-of-ai-human-centered-design-is-more-crucial-than-ever)

* **Better, Easier, Emotional UX** [*Google Design*]
  [https://design.google/library/expressive-material-design-google-research](https://design.google/library/expressive-material-design-google-research)

* **Synthetic Users: If, When, and How to Use AI-Generated “Research”** [*Nielsen*]
  [https://nngroup.com/articles/synthetic-users/](https://nngroup.com/articles/synthetic-users/)

* **Your ‘For You’ Algorithm Disagrees With You** [*Stanford*]
  [https://hai.stanford.edu/news/your-for-you-algorithm-disagrees-with-you](https://hai.stanford.edu/news/your-for-you-algorithm-disagrees-with-you)

#

# Agile Development Fundamentals

Introduction: Agile development is an approach to building products through small, usable increments, frequent feedback, and ongoing adjustment. Teams establish a direction, prioritize the most valuable work, and use what they learn to decide what comes next. This reduces the time between making an assumption and discovering whether it holds up. For product designers and managers, Agile creates opportunities to test ideas before investing too heavily in them. It requires clear goals, close collaboration, and attention to quality. The central question is whether each increment improves the product and informs the next decision.

**The Need to Know**

Product development involves uncertainty. Customers may respond differently than expected, technical constraints may emerge, and business priorities may change. A detailed plan cannot eliminate those conditions. Agile helps teams manage them by shortening the distance between development and feedback. Its foundational principles emphasize early delivery, customer collaboration, technical excellence, sustainable work, and regular reflection.

Several practices make that approach concrete:

| Fundamental | What it means in practice |
| :---- | :---- |
| **Clear Goals and Priorities** | Define the outcome and maintain an ordered backlog of work that could help achieve it |
| **Small, Usable Increments** | Deliver a coherent piece of functionality that can be evaluated before expanding it |
| **Cross-functional Collaboration** | Bring design, engineering, and product expertise together throughout development |
| **Frequent Feedback** | Review working results with stakeholders and observe how intended users respond |
| **Built-in Quality** | Agree on what “done” requires, including relevant testing and acceptance criteria |
| **Continuous Improvement** | Regularly examine how the team works and make specific adjustments |

Agile is an umbrella approach. Scrum provides one structure for applying it through defined roles, events, and timeboxed sprints; other teams organize work around a continuous flow. A two-week schedule or daily meeting does not establish whether a team is learning effectively. Planning, documentation, and technical design still matter. Their depth should reflect the decision and its consequences. For physical products, teams can iterate through prototypes and subsystem tests while respecting manufacturing lead times and validation requirements.

**The Real-world Application**

SpaceX offers a memorable illustration of agile principles through its Starship development program. The company builds and tests successive vehicles, using flight results to identify changes for the next version. After its second integrated flight test in November 2023, SpaceX identified corrective actions that included hardware redesigns and software updates. Testing produced concrete evidence that shaped subsequent development.

On October 13, 2024, Starship’s fifth flight test achieved a major milestone: the returning Super Heavy booster was caught by mechanical arms on the launch tower. This demonstrated one capability needed for the larger goal of a reusable launch system. Teams could evaluate that specific achievement while continuing to develop other parts of the vehicle.

The connection to agile is the discipline of building in increments, testing assumptions, and changing priorities based on evidence. This illustrates agile principles without implying that SpaceX follows a particular Scrum process. For a product team, the practical question is: What can we build and test next that will resolve our most important uncertainty? The value of an unsuccessful test comes from understanding what happened and making a better decision afterward. In hardware, those iterations also have to account for manufacturing costs, safety, and the difficulty of changing something already built.

**How this Applies to You**

Define progress in terms of customer and business outcomes. Completing planned work matters, but a finished feature does not establish that it solved the problem. Before development begins, clarify who the increment serves, what improvement you expect, and how you will evaluate it. Use reviews to examine those assumptions alongside the working product.

Stay involved throughout delivery. Designers should help evaluate implementation details and observe usability as the product takes shape. Product leaders should clarify priorities, resolve tradeoffs, and protect the team from competing demands. Both should involve engineers early enough for technical knowledge to influence the solution. Changes in direction should follow evidence or an explicit business decision, with their costs made visible.

Treat each increment as an opportunity to reduce uncertainty. Sometimes the next step is more functionality; sometimes it is a reliability improvement, additional research, or stopping an idea that did not work. Keep the team focused with a simple question: “What is the smallest useful thing we can deliver that will help us make the next decision?”

**Continue Down the Rabbit Hole …**

* **What Is Agile?** [*McKinsey*]
  [https://mckinsey.com/featured-insights/mckinsey-explainers/what-is-agile](https://mckinsey.com/featured-insights/mckinsey-explainers/what-is-agile)

* **Lean UX & Agile: Study Guide** [*Nielsen*]
  [https://nngroup.com/articles/lean-ux-agile-study-guide/](https://nngroup.com/articles/lean-ux-agile-study-guide/)

* **What Is Agile? An Iterative Approach to Project Management** [*Coursera*]
  [https://coursera.org/articles/what-is-agile-a-beginners-guide](https://coursera.org/articles/what-is-agile-a-beginners-guide)

* **The Future of Project Work** [*Project Management Institute*]
  [https://pmi.org/learning/thought-leadership/future-of-project-work](https://pmi.org/learning/thought-leadership/future-of-project-work)

#

# User Personas & Segmentation

Introduction: User personas and segmentation help product teams understand whom they are serving and which differences should influence the product. Segmentation groups customers or users by meaningful shared characteristics, such as needs, behaviors, experience, or purchasing context. Personas turn research about those patterns into recognizable descriptions of representative users, including their goals, constraints, and frustrations. Together, they help teams make more specific choices about functionality, onboarding, pricing, and support. Their value comes from improving decisions: identifying whose problem matters, understanding the circumstances around it, and designing an experience that addresses it.

**The Need to Know**

“The user” often means something different to everyone in a product meeting. Sales may be thinking about the buyer, engineering about an experienced operator, and design about someone trying the product for the first time. Those people can have competing needs. A purchasing manager might prioritize reporting and cost control, while the employee using the product every day needs speed and simplicity. Making those distinctions explicit helps the team resolve tradeoffs.

Useful personas come from evidence gathered through interviews, observation, surveys, and other research. Their fictional names and profiles summarize real patterns. Include details that affect a decision: technical confidence, time pressure, accessibility needs, or responsibility for a task. Demographic details alone rarely explain what someone needs from a product.

Segmentation should also be actionable. Two groups deserve different treatment when the distinction changes the product, message, service, or business decision. Avoid creating so many segments that each contains too little evidence to evaluate. Remember that context changes: the same person can be an expert in one task, a beginner in another, and a price-sensitive buyer only under certain circumstances.

**The Real-world Application**

Hilton’s hotel portfolio shows how different travel needs translate into different offerings. Its brands span luxury experiences, locally inspired stays, spacious suites, and essential accommodations. Homewood Suites specifically emphasizes longer stays, offering studio and larger suites with kitchens that include refrigerators, dishwashers, microwaves, and stovetops. These are concrete product differences tied to the circumstances of a stay.

Viewed through segmentation, length of stay and desired experience provide useful ways to distinguish demand. A guest staying for several weeks may place more value on preparing meals and maintaining a routine than someone visiting for a single night.

A team designing a similar extended-stay experience could investigate a persona such as “the professional temporarily relocating for work.” Research would need to establish their actual routines, budget constraints, and frustrations. If supported by evidence, that persona could inform kitchen storage, workspace layout, booking information, and housekeeping options. The segment identifies a group worth serving; the persona helps the team work through the details of the experience.

**How this Applies to You**

Be precise when discussing customer needs. Replace broad claims such as “people want more options” with a description of who needs those options, under what conditions, and why. Distinguish the buyer, administrator, and end user when they are different people. Then decide which needs the product will prioritize and explain the consequences of that choice.

Treat personas and segments as working hypotheses that require maintenance. Recruit research participants across relevant groups, include people who abandoned the product, and check whether observed behavior supports your assumptions. A persona can guide a design discussion, but it cannot answer a research question on a customer’s behalf.

Measure whether your choices help the intended audience. An overall increase in adoption might hide a decline among the users you most need to serve. Compare outcomes by meaningful segment and update your understanding when patterns change. The question to keep asking is: “For whom is this the right experience, and what evidence supports that?”

**Continue Down the Rabbit Hole …**

* **5 Customer Segmentation Methods to Understand Purchasers** [*Amplitude*]
  [https://amplitude.com/blog/customer-segmentation/](https://amplitude.com/blog/customer-segmentation/)

* **Personas Make Users Memorable** [*Nielsen*]
  [https://nngroup.com/articles/persona/](https://nngroup.com/articles/persona/)

* **Bridging Personas and Product Design for Creative Ideation** [*Toyota Research Institute and Northwestern University researchers*]
  [https://arxiv.org/abs/2602.06197](https://arxiv.org/abs/2602.06197)

* **Customer Segmentation: Targeted Marketing Made Simple** [*Adobe*]
  [https://business.adobe.com/blog/basics/define-customer-segmentation](https://business.adobe.com/blog/basics/define-customer-segmentation)

#

# Jobs-to-be-Done [Framework]

Introduction: People don't buy products simply because they fit a demographic. They “hire” them to make progress in a specific situation. Jobs to Be Done, or JTBD, helps us understand what someone is trying to accomplish, what prompted them to look for a solution, and why they chose one alternative over another. In Predictable Product Development, it connects customer research to decisions about what to build and how to position it. Understanding the job gives us a clearer definition of the market, the competition, and the outcome our product needs to deliver.

**The Need to Know**

Knowing someone’s age, income, or job title can help you find potential customers. It rarely explains the full reason they buy. Two people with similar backgrounds may need very different things, while people with little in common may be trying to make the same progress. JTBD focuses on those circumstances and recognizes that a decision can have functional, emotional, and social dimensions: what someone needs to accomplish, how they want to feel, and how they want others to see them.

The framework also changes your competitive analysis. If the job is to coordinate a small team’s work, a project management app might compete with email, a spreadsheet, a weekly meeting, or simply remembering who promised to do what. Customers compare your product with whatever they currently use to make progress. That means a technically superior product can lose when adoption requires more effort than the improvement is worth.

**The Real-world Application**

**Apple’s** original iPhone is one of the most famous (Steve) Jobs-to-be-done case studies. In its 2007 announcement, Apple presented a single device combining a mobile phone, a widescreen iPod, and internet capabilities including email, browsing, and maps. The offering brought several previously separate activities into one handheld product.

Through a JTBD lens, one possible job was: “When I’m away from home or my desk, help me stay connected, find information, and occupy my downtime without juggling several devices.” It shows how defining the job around a situation can reveal competition across product categories. A team investigating this job would study the effort of carrying devices, switching between them, and accessing information on the move, alongside the individual capabilities of each device.

Consider customers who recently adopted, replaced, or stopped using a relevant solution. Reconstruct the decision: what happened first, what they tried, what frustrated them, what attracted them to the alternative, and what nearly stopped them from switching. Summarize the pattern with a statement such as, “When [situation], I want to [make progress], so I can [achieve an outcome].” Treat that statement as a research finding to test across additional customers; describe a need independently of your proposed feature.

**How this Applies to You**

It is our opportunity to become more curious about the circumstances behind demand. A customer asking for a feature is giving you a starting point. Find out what happened that made the feature matter, how they handle the situation today, and what a better outcome would look like. This helps you distinguish a useful improvement from a request that sounds important but has little effect on the customer’s progress.

You also need to account for the difficulty of switching. Customers may dislike their current approach and still keep using it because it is familiar, their team understands it, or they worry about losing something during the transition. Product work may therefore include migration, onboarding, reassurance, and compatibility. These can determine whether someone ever experiences the value of the feature you built.

Before adding something to the roadmap, ask: What progress is the customer hiring us to help them make, and what are we replacing? Then measure whether they achieve that progress. For the manager preparing a budget presentation, generating more reports is a weak measure of success. Being able to explain the spending accurately, with less preparation and fewer unanswered questions, is much closer to the job.

**Continue Down the Rabbit Hole …**

* **The Jobs to Be Done Framework Explained** [*Harvard Business School*]
  [https://online.hbs.edu/blog/post/jobs-to-be-done-examples](https://online.hbs.edu/blog/post/jobs-to-be-done-examples)

* **The Ultimate Guide to JTBD with Bob Moesta** [*Lenny’s Newsletter*]
  [https://lennysnewsletter.com/p/the-ultimate-guide-to-jtbd-bob-moesta](https://lennysnewsletter.com/p/the-ultimate-guide-to-jtbd-bob-moesta)

* **Transforming the Voice of the Customer:** Large Language Models for Identifying Customer Needs [*Northwestern Kellogg and MIT researchers*]
  [https://arxiv.org/abs/2503.01870](https://arxiv.org/abs/2503.01870)

* **Jobs to Be Done 101:** Your Interviewing Style Primer [*Dscout*]
  [https://dscout.com/people-nerds/the-jobs-to-be-done-interviewing-style-understanding-who-users-are-trying-to-become](https://dscout.com/people-nerds/the-jobs-to-be-done-interviewing-style-understanding-who-users-are-trying-to-become)

Jobs To Be Done Template(s) by *Figma* A customizable FigJam template for exploring why customers “hire” your product, what barriers they face, and how well it meets their needs.[https://figma.com/templates/jobs-to-be-done-examples/](https://figma.com/templates/jobs-to-be-done-examples/)

#

# Innovation Adoption Curve

Introduction: The Innovation Adoption Curve describes how people adopt a new idea or product at different times. Associated with Everett Rogers’ Diffusion of Innovations, it groups adopters into innovators, early adopters, early majority, late majority, and laggards. Customers willing to try your first version may have different needs from those who buy later; early enthusiasm is useful evidence, but reaching a broader market may require changes to the product, price, onboarding, support, and proof of value.

**The Need to Know**

Early customers can make a product feel more ready than it is. They may tolerate complicated setup, missing features, or frequent troubleshooting because they see an advantage worth pursuing. A broader audience may expect the product to work immediately and fit existing routines. If your research only includes people already excited about the idea, you can miss the reasons everyone else is waiting.

The framework provides a useful starting point for thinking about those differences:

| Adopter Group | Adoption Pattern | Product Implication |
| :---- | :---- | :---- |
| **Innovators** | Explore new possibilities and accept substantial uncertainty | Make experimentation possible |
| **Early Adopters** | Recognize a useful advantage and can influence others | Demonstrate value in a specific use case |
| **Early Majority** | Deliberate before adopting and look for evidence | Provide reliable performance and credible references |
| **Late Majority** | Adopt when uncertainty has fallen and adoption is widespread | Reduce cost, complexity, and switching effort |
| **Laggards** | Adopt last, often as existing approaches remain preferable or resources are limited | Understand the practical reasons for waiting |

These categories describe relative adoption timing, not permanent personality types. Someone can adopt one technology early and another late. The familiar percentages shown on the curve are model conventions, not guaranteed market shares or a sales forecast.

For product teams, the opportunity is to identify what makes adoption easier: a clear advantage, compatibility with existing habits, manageable complexity, a chance to try the product, and visible results. A demonstration or simpler setup process may do more for adoption than another advanced feature. The curve helps you ask what the next group of customers needs before it is willing to commit.

**The Real-world Application**

**Nintendo’s Wii** illustrates how product design can broaden adoption beyond an established enthusiast audience. In its 2007 financial review, Nintendo described a strategy of expanding gaming to more people. It reported that the Wii’s intuitive remote-based interaction attracted both conventional gamers and additional audiences. The console sold 5.84 million units worldwide within its first five months.

Through the adoption framework, Wii Sports helps explain the approach. Familiar activities such as tennis and bowling gave newcomers an understandable starting point. Watching someone play made the interaction easier to grasp, and trying it together reduced the effort of learning alone.

The lesson is that broader adoption can require changing the entry experience. Nintendo gave people who were less familiar with gaming a reason and a way to participate. For your product, consider what a newcomer must understand before experiencing value. If that requires extensive explanation from an enthusiastic early user, the product may still have an adoption barrier to solve.

**How this Applies to You**

You should be explicit about whose readiness you are measuring. Recruiting enthusiastic early users can help you discover promising uses and improve an initial product. Research with hesitant customers can reveal different issues: unclear benefits, switching costs, lack of trust, or a perfectly adequate existing solution. Both groups provide useful evidence, but they answer different questions.

Designers should investigate where the experience assumes knowledge or motivation that later customers may not have. Product owners should examine the complete adoption decision, including purchase, migration, training, support, and ongoing cost. Avoid assuming that additional marketing will resolve a problem rooted in the product or the effort required to use it.

**Continue Down the Rabbit Hole …**

* **Innovation Adoption by Forward-Looking Social Learners** [*Theoretical Economics*]
  [https://onlinelibrary.wiley.com/doi/10.3982/TE4455](https://onlinelibrary.wiley.com/doi/10.3982/TE4455)

* **Electric Vehicles Are on the Road to Mass Adoption** [*RMI*]
  [https://rmi.org/resources/electric-vehicles-are-on-the-road-to-mass-adoption/](https://rmi.org/resources/electric-vehicles-are-on-the-road-to-mass-adoption/)

* **The State of Generative AI Adoption** [*Federal Reserve Bank*]
  [https://stlouisfed.org/on-the-economy/2025/nov/state-generative-ai-adoption-2025](https://stlouisfed.org/on-the-economy/2025/nov/state-generative-ai-adoption-2025)

* **Precision Agriculture: Benefits and Challenges for Technology Adoption and Use** [*U.S. Government Accountability Office*]
  [https://gao.gov/products/gao-24-105962](https://gao.gov/products/gao-24-105962)

# Behavioral Economics

Introduction: People don't evaluate every product decision by carefully weighing all the costs and benefits. We rely on habits, take shortcuts, respond to how choices are presented, and put things off even when we intend to act. Behavioral economics combines psychology and economics to explain these patterns. In Predictable Product Development, it helps us understand why a useful product might go unused and why a small change in timing, wording, or effort can affect adoption. The goal is to design experiences that account for how people actually make decisions.

**The Need to Know**

A customer can understand a product’s value and still never get around to using it. The benefit may feel distant, the setup may require immediate effort, or the existing approach may feel safer simply because it is familiar. Behavioral economics helps explain the gap between intention and action. Research on limited rationality and self-control shows why information and incentives alone may be insufficient to change behavior.

Several concepts are particularly useful in product work. Present bias helps explain why an immediate inconvenience can outweigh a future benefit. Loss aversion describes how a potential loss can matter more than an equivalent gain. Defaults influence decisions because people may stick with a preselected option, while framing changes how they interpret the same underlying choice. For example, a customer evaluating new software may focus heavily on the disruption of switching, even when they expect it to save time later. These concepts suggest explanations to investigate, rather than guaranteed formulas for changing behavior.

The practical implication is that presentation and process are part of the product’s value. A complicated comparison page, an unclear next step, or a poorly timed request can prevent someone from acting. Before adding another incentive, investigate what makes the desired action difficult. Sometimes the better investment is reducing effort, making consequences easier to understand, or asking at a moment when the customer is ready.

**The Real-world Application**

**Bank of America’s** Keep the Change program launched in 2005, allowing enrolled customers to round debit card purchases up to the next dollar and transfer the difference from checking into savings. A $4.60 purchase, for example, triggers a $0.40 transfer. Once customers enroll, saving becomes part of a routine they already perform, without requiring a separate decision after each purchase.

Viewed through behavioral economics, the design reduces the effort of following through on a savings intention. It also fits mental accounting, the tendency to treat money differently depending on how we categorize it. Setting aside “spare change” may feel more manageable than making a deliberate transfer, even though the money comes from the same person’s account. Money moved into a savings account also doesn't, by itself, establish that someone’s total savings increased.

**How this Applies to You**

Behavioral economics is usually taught as removing friction, but the useful version is placing it deliberately. Amazon’s one-click patent was about removing friction from purchase. A two-step confirmation before deleting a workspace is about adding it where the cost of a mistake is high. The test is whose interest the friction serves: a confirmation that protects the user is good design, and a cancellation flow with six screens and a retention offer is the same technique pointed the other way.

**Continue Down the Rabbit Hole …**

* **The Hidden Why: Behavioral Economics for UX** [*Nielsen*]
  [https://nngroup.com/articles/behavioral-economics-for-ux/](https://nngroup.com/articles/behavioral-economics-for-ux/)

* **How We Built the Friend Streak on Duolingo** [*Duolingo*]
  [https://blog.duolingo.com/product-lessons-friend-streak/](https://blog.duolingo.com/product-lessons-friend-streak/)

* **Behavioral Nudges Prevent Loan Delinquencies at Scale** [*PNAS*]
  [https://doi.org/10.1073/pnas.2416708122](https://doi.org/10.1073/pnas.2416708122)

* **Wealthier Workers Benefit Most from Retirement Savings “Nudges”** [*MIT Sloan*]
  [https://mitsloan.mit.edu/ideas-made-to-matter/wealthier-workers-benefit-most-retirement-savings-nudges](https://mitsloan.mit.edu/ideas-made-to-matter/wealthier-workers-benefit-most-retirement-savings-nudges)

#

# Cognitive Biases

Introduction: Cognitive biases are predictable patterns in how we interpret information and make judgments that can lead us to the wrong conclusions. In product development, they influence which customer problems we notice, which ideas we support, and what we accept as evidence that something works. Being smart, experienced, or informed does not make your judgment objective; you might favor a concept because you created it, hear what you hoped to hear in a customer interview, or keep funding a feature because your team has already spent months building it.

**The Need to Know**

Product designers and managers constantly make decisions with incomplete information. You have limited research, competing priorities, and a deadline. Mental shortcuts help you move forward, but they can also make weak evidence feel convincing. A memorable customer complaint can start to look like everyone’s problem. An executive’s estimate can become the team’s starting point without anyone checking it. Once people commit to a direction, they may become better at defending it than evaluating it.

Several biases are especially useful to recognize in product work:

| Bias | Description |
| :---- | :---- |
| **Anchoring** | Letting the first estimate or proposed price shape the discussion too heavily |
| **Resulting** | Judging a product decision by its outcome rather than the quality of the reasoning and information available when it was made. |
| **Hindsight Bias** | Once you know the answer, it feels like you always knew it. This quietly destroys learning, because a team that believes the failure was obvious never examines what it actually knew at the time. Write predictions down beforehand and the bias has nowhere to hide |
| **Sunk Cost Fallacy** | Continuing to invest because of what you have already spent. The six months in the ground are gone whichever way you decide; the only honest question is whether you would start this project today, knowing what you now know |
| **Status Quo Bias** | Preferring things to stay as they are, and treating the current state as neutral when it isn't. It's why the migration keeps slipping and why the default option wins — which also makes it a lever, not only a trap |
| **Cognitive Dissonance** | The discomfort of holding two conflicting beliefs, usually resolved by quietly discarding the inconvenient one. In practice it looks like a team explaining away the research that contradicts the roadmap it has already committed to |
| **Action Bias** | The pull toward doing something rather than waiting, especially while being watched. Goalkeepers dive on penalty kicks because standing still looks like negligence, even though standing still saves more of them. Roadmaps fill up for the same reason |
| **Confirmation Bias** | Seeking and over-weighting evidence that supports what you already believe. It's why user interviews so often confirm the plan: the questions were leading, the sample was friendly, and the transcript got read for quotes rather than for surprises |
| **Survivorship Bias** | Studying active customers while overlooking people who abandoned the product |

These patterns affect both the team building the product and the people using it. However, calling something a bias should never become a convenient way to dismiss customer feedback. Someone might resist a redesign because it genuinely makes their work harder. Your job is to investigate the reason, including explanations that contradict your own.

**The Real-world Application**

**Coca-Cola’s** introduction of New Coke in 1985 shows how extensive research can still miss a critical part of a product decision. According to the company’s history, the new formula was preferred in taste tests involving nearly 200,000 consumers. But the tests did not capture people’s emotional attachment to the original product. After a strong backlash, Coca-Cola brought back the original formula 79 days after introducing the replacement.

Viewed through the lens of cognitive biases, the case raises questions about how teams frame a problem and interpret favorable evidence. Preferring a taste in a test does not necessarily mean wanting a familiar product replaced. The episode does not prove that individual executives had a particular bias. It does illustrate the risk of treating evidence about one attribute as sufficient evidence for a much broader decision.

For a product team, the application is straightforward: test the actual decision customers will experience. If you are replacing an existing product, research should examine what users would lose, how their routines would change, and whether they would choose the replacement in context. Before committing, ask someone to make the strongest evidence-based case against your recommendation.

**How this Applies to You**

Treat your own judgment as something to test. Before research begins, write down your assumptions, alternative explanations, and the results that would make you change direction. Ask customers about actual experiences without signaling your preferred answer. When reviewing findings, separate what you observed from what you inferred. “Three participants struggled to find the button” is an observation. “We need a complete redesign” is a proposed response.

Build independent thinking into team decisions. Have people evaluate concepts before hearing the most senior person’s opinion. Compare estimates with similar completed projects. Establish criteria for continuing, changing, or stopping work before the team becomes deeply invested. These habits give people a practical way to challenge a decision without turning the discussion into a contest of confidence. The question to keep asking is: “What would we need to learn to change our minds?”

**Continue Down the Rabbit Hole …**

* **The science of navigating business uncertainty** [*Bocconi School of Management*]
  [https://sdabocconi.it/en/sda-bocconi-insight/knowledge/the-science-of-navigating-business-uncertainty/](https://sdabocconi.it/en/sda-bocconi-insight/knowledge/the-science-of-navigating-business-uncertainty/)

* **Bias in the Loop: How Humans Evaluate AI-Generated Suggestions** [*Harvard*]
  [https://hdsr.mitpress.mit.edu/pub/nrcn4h7d](https://hdsr.mitpress.mit.edu/pub/nrcn4h7d)

* **The Impact of Generative AI on Critical Thinking** [*Microsoft*]
  [https://microsoft.com/en-us/research/publication/the-impact-of-generative-ai-on-critical-thinking-self-reported-reductions-in-cognitive-effort-and-confidence-effects-from-a-survey-of-knowledge-workers/](https://microsoft.com/en-us/research/publication/the-impact-of-generative-ai-on-critical-thinking-self-reported-reductions-in-cognitive-effort-and-confidence-effects-from-a-survey-of-knowledge-workers/)

* **Review of Use of Dark Patterns Affecting Subscription Services, Privacy** [*FTC*]
  [https://ftc.gov/news-events/news/press-releases/2024/07/ftc-icpen-gpen-announce-results-review-use-dark-patterns-affecting-subscription-services-privacy](https://ftc.gov/news-events/news/press-releases/2024/07/ftc-icpen-gpen-announce-results-review-use-dark-patterns-affecting-subscription-services-privacy)

# Maslow's Hierarchy of Needs

Introduction: People’s needs range from basic survival and safety to belonging, esteem, and personal growth. Maslow’s Hierarchy of Needs offers a lens for exploring these motivations in product development; helping teams understand customer priorities and design experiences that support both essential needs and deeper aspirations.

**The Need to Know**

Maslow described five groups of needs: physiological needs, such as food and sleep; safety, including security and stability; love and belonging, including relationships and acceptance; esteem, including confidence, achievement, and respect; and self-actualization, or realizing one’s potential. The familiar hierarchy suggests that unmet basic needs often take priority. But it isn't a rigid sequence. People can pursue several needs at once, and Maslow himself acknowledged exceptions to their ordering. Use it as a framework for asking better questions, not a formula for predicting every customer’s behavior.

For product teams, this helps explain why two people can use the same product for very different reasons. Someone might join a gym to improve their health, spend time with friends, feel more confident, or prepare for a personal challenge. The equipment is the same. The motivation changes what matters in the experience, what keeps the person coming back, and what alternatives they consider.

These motivations also show up in business software. A manager evaluating a reporting tool may want faster access to numbers, but they may also worry about presenting incorrect information to their boss. That gives the product team a hypothesis to investigate: perhaps traceable calculations and clear explanations matter more than another dashboard. You still need customer research to determine whether that is true. The framework helps you know what to look for.

**The Real-world Application**

Airbnb is a useful case study. Its basic offering gives travelers a place to sleep. But in its 2014 redesign, Airbnb emphasized hosts, personal relationships, and neighborhood context, alongside a new identity centered on belonging. The experience and positioning gave customers more to consider than a property's location, amenities, and price.

Through Maslow’s framework, we can interpret that offering at several levels. A comfortable place to sleep addresses a physical need. Confidence in the accommodation addresses safety. Feeling welcomed by a host or connected to a neighborhood can support belonging. For some travelers, exploring an unfamiliar place might also support personal growth. The product lesson is practical: a promise of connection becomes harder to deliver when the guest is worried about whether they can get into the apartment.

**How this Applies to You**

As a product leader, part of your job is to understand why a task matters to the person doing it. A request to “make this easier” can mean several things: save me time, help me avoid embarrassment, reduce my uncertainty, or let me do this without asking someone else for help. Each explanation points toward a different design decision. Your responsibility is to investigate which one applies.

Be careful about assuming that a more emotional or aspirational benefit is automatically more valuable. A dependable product that helps someone feel secure can be enormously useful. Adding a community, achievement badges, or inspirational messaging only makes sense when it supports something customers actually want. Where a need appears in the hierarchy doesn't determine its commercial value or its place on your roadmap.

Before committing to a solution, ask: What does getting this done make possible for this person? Then keep asking until you have an answer specific enough to guide a product decision. That is where a general theory of motivation becomes useful design work.

**Continue Down the Rabbit Hole …**

* **What Is the Hierarchy of Needs?** [*Interaction Design Foundation*]
  [https://ixdf.org/literature/topics/hierarchy-of-needs](https://ixdf.org/literature/topics/hierarchy-of-needs)

* **Turning Maslow’s Hierarchy on Its Head** [*IDEO*]
  [https://ideo.org/perspective/turning-maslows-hierarchy-on-its-head](https://www.ideo.org/perspective/turning-maslows-hierarchy-on-its-head)

* **Maslow’s Need Theory in the Age of Artificial Intelligence** [*Elsevier Journal*]
  [https://doi.org/10.1016/j.techfore.2025.124222](https://doi.org/10.1016/j.techfore.2025.124222)

* **Beyond Maslow’s Pyramid** [*TU Delft Researchers*]
  [https://research.tudelft.nl/en/publications/beyond-maslows-pyramid-introducing-a-typology-of-thirteen-fundame/](https://research.tudelft.nl/en/publications/beyond-maslows-pyramid-introducing-a-typology-of-thirteen-fundame/)

#

# Prediction Mechanisms

Introduction: Every product decision includes a prediction. We believe customers will want something, understand how to use it, pay for it, or change their behavior because of it. In this course, prediction mechanisms are the methods we use to make those expectations explicit, gather evidence, and improve our judgment before making a larger investment. These include prototypes, experiments, forecasts, and comparisons with similar products or past decisions. The goal is to reduce uncertainty enough to make a sound decision, then compare what actually happened with what we expected.

**The Need to Know**

A roadmap is full of predictions, even when nobody calls them that. “This feature will improve retention” is a prediction. So is “customers will pay more for this” or “removing this step will increase conversion.” Left unexamined, these statements can become commitments simply because someone said them confidently. Making the prediction explicit gives the team something to investigate before spending months delivering it.

Different questions require different mechanisms. A prototype can help you assess whether people understand an interaction. A paid pilot can provide evidence that a customer will commit money and effort. Historical results from comparable launches can inform a forecast. A properly designed randomized experiment can help estimate whether a change caused an improvement. Each method answers a particular question. Someone successfully using a prototype doesn't establish that they will buy the product, and a purchase doesn't establish that they will keep using it.

Specificity makes the prediction useful. “Improve onboarding” is an intention. “Increase the share of new customers who complete setup within seven days from 40% to 50%” is a testable prediction. The team can identify the target customers, define completion, establish a baseline, and decide what evidence would justify a wider release. The numbers in this example are illustrative, but the discipline applies to any product: define the expected change before looking at the results.

**The Real-world Application**

In its published account of its experimentation platform, **Netflix** describes testing changes to its interface, recommendations, and streaming technology against a control experience. It also tests the artwork used to present titles. Its team reported that some artwork tests produced 20% to 30% more viewing for the affected title. The design decision had a measurable behavioral consequence that the team could evaluate before making it standard.

Netflix has also described personalizing artwork based on what might appeal to individual members. This illustrates two related uses of prediction: a model estimates which image will be relevant to a person, while an experiment evaluates whether using that approach improves the experience. A model can make plausible predictions without producing a better product, so the product decision still needs testing.

**How this Applies to You**

We must be able to explain the expected consequence of a product design choice. That means connecting what you want to build with a change in customer behavior or a business outcome. Your reasoning should include the evidence you have, the assumptions you are making, and the uncertainty that remains. “We don't know yet” is a useful answer when it leads to a concrete learning plan.

You also need to distinguish a good decision from a fortunate result. A weak idea can succeed because of timing, and a well-supported idea can fail because conditions change. Keep a record of important predictions and revisit them. Over time, this helps you notice whether you consistently overestimate adoption, underestimate switching costs, or mistake initial interest for sustained demand. Your past judgment becomes something you can learn from.

Before approving the next feature or experiment, ask: What are we predicting, what evidence would change our minds, and when will we know? Answering those questions gives the team a shared basis for making the decision and a clear reason to revisit it.

**Continue Down the Rabbit Hole …**

* **How Big Things Get Done**: The surprising factors that determine the fate of every project. [Penguin Random House]
  [https://www.penguinrandomhouse.com/books/672118/how-big-things-get-done-by-bent-flyvbjerg-and-dan-gardner/](https://www.penguinrandomhouse.com/books/672118/how-big-things-get-done-by-bent-flyvbjerg-and-dan-gardner/)

* **Testing Business Ideas:** How to increase the success of any venture through rapid experimentation [*Strategyzer*]
  [https://strategyzer.com/library/testing-business-ideas-book](https://strategyzer.com/library/testing-business-ideas-book)

* **Superforecasting:** Methods used to assess probabilistic predictions. [*Wharton*]
  [https://www.youtube.com/watch?v=6POQjSjIXWk](https://www.youtube.com/watch?v=6POQjSjIXWk)

* **Trustworthy Online Controlled Experiments:** A Practical Guide to A/B Testing [*Cambridge University*]
  [https://experimentguide.com/](https://experimentguide.com/)

# Power Law Distribution

Introduction: Product outcomes can be extremely uneven. A small number of products, features, or investments may account for a disproportionate share of the value created. A power law is a specific mathematical distribution that can describe this pattern, where most outcomes are small and a few are exceptionally large. In Predictable Product Development, it challenges the assumption that every initiative deserves similar resources or will produce similar results. It encourages us to investigate where value concentrates, test opportunities before making large commitments, and recognize when one successful bet could materially change the business.

**The Need to Know**

With a familiar bell curve, observations cluster around an average and extreme outcomes become very rare. In a power law distribution, large outcomes remain more likely than a bell curve would suggest. More precisely, increasing the outcome threshold by a fixed multiple reduces the probability of exceeding it by a consistent factor across the range where the power law holds. The practical implication is that an average can hide enormous differences between typical results and exceptional ones.

For product teams, this changes how you think about prioritization. Ten projects of similar cost might produce wildly different benefits. One workflow improvement could remove the main reason customers leave, while several other features barely affect their experience. You need to understand the distribution of outcomes before assuming that spreading resources evenly is sensible. You also need room to explore, because the biggest opportunities aren't always obvious in advance.

Be precise about the terminology. The 80/20 rule is a useful shorthand for concentration, but it isn't a universal ratio or proof of a power law. A product with a few highly active users doesn't automatically have a power law distribution. Other distributions can produce similarly uneven results, and establishing a power law requires statistical testing. For this course, the immediate lesson is to examine concentration in your data while being careful about the mathematical claims you make.

**The Real-world Application**

**Amazon** is one of the most powerful examples of managing for unusually large outcomes. In his 2015 shareholder letter, Jeff Bezos explained that a major success can pay for many unsuccessful experiments. He identified AWS, Marketplace, and Prime as bold bets that became substantial businesses. He also described Amazon’s unsuccessful Auctions and zShops efforts before Marketplace. The letter illustrates a strategy built around uneven returns, though it doesn't establish that Amazon’s results follow a statistically verified power law.

For a product team, the useful real-world example is to consider both the probability of success and the size of the possible outcome. A modest improvement with a high chance of working and a more uncertain opportunity with substantial upside serve different purposes. You can fund dependable improvements while reserving capacity for experiments that might reveal a much larger opportunity. That only works if you can afford the unsuccessful experiments and have a credible reason to believe the upside exists.

Consider an outcome such as repeat usage, revenue, or time saved. Examine how much of it comes from different customer segments, workflows, or use cases. Then separate what you observe today from what you believe could create value tomorrow. Design a small test around the most promising opportunity and specify what evidence would justify more investment. Low usage alone is insufficient reason to remove something: account recovery, for example, may be rarely used but essential when needed.

**How this Applies to You**

We must be comfortable recommending unequal investment when the evidence supports it. A roadmap with something for everyone may look balanced while failing to meaningfully improve the product. Your responsibility is to explain where additional effort is likely to matter most, including the customer needs and operational work that support your strongest opportunities.

You also need to avoid judging ideas solely by how often they succeed. A team can achieve a high success rate by choosing only small, predictable improvements and still miss a major opportunity. Equally, repeated failure doesn't prove that a breakthrough is approaching. Each experiment needs a reason to exist, a manageable cost, and evidence that helps determine what happens next.

The question to carry forward is: Where could a successful outcome create disproportionate value, and what would we need to learn before committing more resources? This connects ambition with disciplined product work. You give promising ideas enough room to develop while protecting the team’s ability to keep learning.

**Continue Down the Rabbit Hole …**

* **Power-Law Distributions in Empirical Data** [*SIAM Review*]
  [https://arxiv.org/abs/0706.1062](https://arxiv.org/abs/0706.1062)

* **2015 Letter to Shareholders** [*Jeff Bezos, Amazon*]
  [https://ir.aboutamazon.com/files/doc_financials/annual/2015-Letter-to-Shareholders.PDF](https://ir.aboutamazon.com/files/doc_financials/annual/2015-Letter-to-Shareholders.PDF)

* **Black Swan Farming** [*Y Combinator, Paul Graham*]
  [https://paulgraham.com/swan.html](https://paulgraham.com/swan.html)

* **Tails, You Win** [*Collaborative Fund*]
  [https://collabfund.com/blog/tails-you-win/](https://collabfund.com/blog/tails-you-win/)

# J Curve [Portfolio Theory]

Introduction: The J Curve describes an investment pattern in which returns are initially negative before improving over time, creating a shape that resembles the letter “J.” In product development, the idea helps explain why a promising initiative may consume money and resources before generating meaningful returns. Research, development, infrastructure, and customer acquisition often come before revenue. From a portfolio perspective, the challenge is to balance investments at different stages so the business can support future growth while meeting its current needs. The upward turn is never guaranteed. Teams need evidence that continued investment is improving the likelihood of a worthwhile return.

**The Need to Know**

The J Curve is commonly associated with private equity, where early fees and expenses can produce negative returns before investments mature and gains emerge. Applied to product portfolios, it offers a way to think about the timing of investment and payoff. A new product may need substantial funding before customers can even try it. Evaluating that initiative against a mature product’s current profitability can lead a business to abandon promising opportunities too early.

The opposite mistake is equally costly: treating every loss as evidence that success is coming. Spending money does not automatically create value, and time alone does not fix weak demand or poor economics. Product teams need to distinguish an expected investment period from a deteriorating business case. Repeat usage, willingness to pay, falling delivery costs, and improving margins can provide evidence that an initiative is progressing.

| Portfolio consideration | Question to ask |
| :---- | :---- |
| **Depth of the dip** | How much funding could this initiative require before it starts supporting itself? |
| **Duration of the dip** | How long can we fund it, including potential delays? |
| **Evidence of progress** | What should improve before we commit more resources? |
| **Potential payoff** | Is the expected return worth the remaining cost and risk? |
| **Timing across investments** | Can the business support several initiatives needing funding simultaneously? |

At the portfolio level, timing matters as much as the attractiveness of individual ideas. Several promising products can create a cash shortage if they all require heavy investment at once. Established products may help fund newer initiatives, while smaller experiments let the business test opportunities before making larger commitments.

**The Real-world Application**

**Walmart’s** U.S. eCommerce business offers a useful illustration of delayed profitability within a larger business portfolio. In its first-quarter fiscal 2026 results, announced in May 2025, Walmart reported that U.S. eCommerce had achieved profitability for the first time. The same report showed U.S. eCommerce sales growing 21%. This provides a concrete example of an initiative reaching a profitability milestone after an extended period of development and operation.

Viewed through the J Curve, the case highlights the challenge of funding an expanding business before it becomes profitable. For a product team building an online retail offering, the investment case would need to account for ordering software, fulfillment, delivery, and customer support. Growing sales would be one signal of progress, but the team would also need to understand whether serving each order was becoming more economical.

The distinction between current profitability and recovering the original investment matters here. A profitable quarter does not establish that all prior spending has been earned back. Teams should track both whether an initiative can sustain its ongoing operations and whether its expected future returns justify the total investment.

**How this Applies to You**

You should be able to explain the investment logic behind your roadmap. What needs to be built before value can emerge? Which assumptions make the expected payoff credible? How much time and money will it take to test those assumptions? Designers influence this through choices about scope, usability, and technical complexity. Product leaders connect those choices to demand, resources, and business outcomes.

Commit resources in stages, with clear expectations for what each stage should establish. An early prototype might test whether customers understand and want the offering. A pilot might test repeat usage and delivery costs. A broader launch might test whether the economics hold as adoption grows. Agree on the evidence that would justify expanding, changing, or stopping the initiative before the team becomes attached to it.

Finally, judge the next investment on its expected future value. Money already spent cannot justify spending more. The J Curve should help you give promising products enough time to develop while remaining willing to stop when the evidence no longer supports the opportunity.

**Continue Down the Rabbit Hole …**

* **J-curve: Definition, drivers and mitigation strategies** [*Carta*]
  [https://carta.com/learn/private-funds/management/fund-performance/j-curve/](https://carta.com/learn/private-funds/management/fund-performance/j-curve/)

* **The Productivity J-Curve** [*American Economic Journal*]
  [https://aeaweb.org/articles?id=10.1257%2Fmac.20180386](https://aeaweb.org/articles?id=10.1257%2Fmac.20180386)

* **The state of AI in 2026: On the road to ROI** [*McKinsey*]
  [https://mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai](https://mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai)

* **The AI productivity take-off is finally visible** [*Financial Times*]
  [https://ft.com/content/4b51d0b4-bbfe-4f05-b50a-1d485d419dc5](https://ft.com/content/4b51d0b4-bbfe-4f05-b50a-1d485d419dc5)

# Barbell Theory [Blue Ocean Strategy]

Introduction: A barbell approach concentrates resources at two ends: protecting a dependable business and making smaller, controlled investments in opportunities with substantial upside. Applied to product development, this means maintaining the products customers already rely on while reserving capacity to explore fundamentally different ways to create value. It can support Blue Ocean Strategy by funding experiments that reach underserved customers or create new demand. The concepts serve different purposes: the barbell guides how you allocate resources and exposure to risk, while Blue Ocean Strategy guides the search for differentiated customer value at a lower cost.

**The Need to Know**

Product teams face pressure to deliver immediate results while also finding future sources of growth. Existing customers need improvements, sales teams need commitments fulfilled, and the business needs dependable revenue. Those demands can consume the entire roadmap. A barbell approach makes exploration an explicit investment, with resources that survive beyond the next urgent request.

The underlying logic is to protect against losses that could threaten the business while retaining opportunities for unusually large gains. In product terms, an experiment can pursue an ambitious outcome while keeping the initial financial commitment small. This adapts the barbell approach associated with Nassim Nicholas Taleb, which combines highly conservative and speculative positions while avoiding excessive exposure to the middle. There is no universal allocation that every product organization should follow.

Blue Ocean Strategy gives the exploratory side a direction. It asks how a business can create new demand through value innovation, pursuing differentiation and low cost together. Simply entering another market or adding advanced technology does not establish a blue ocean. Teams need to understand what prevents people from participating today and which expensive industry conventions could be removed.

| Portfolio Choice | Purpose | Useful evidence |
| :---- | :---- | :---- |
| **Protect and improve the core** | Sustain customer value and the business’s ability to operate | Retention, reliability, margins, and customer outcomes |
| **Fund bounded experiments** | Test opportunities beyond the established offering | Unmet needs, willingness to try or pay, and feasible delivery costs |
| **Increase investment selectively** | Expand opportunities that demonstrate promise | Repeat demand, workable economics, and a credible path to scale |

**The Real-world Application**

**3M** provides a useful example of protecting room for exploration within an established business. Its documented “15% Culture” allows employees to spend a portion of their working time pursuing projects they choose in support of business objectives. This gives ideas a route into development even when they fall outside employees’ immediate assignments.

Post-it Notes illustrate what that exploratory capacity can produce. The product emerged from the combination of Spencer Silver’s unusual adhesive and Art Fry’s search for a bookmark that would stay in place. The resulting removable notes created a practical way to leave messages and organize information. 3M introduced Post-it Notes in 1980 and describes the launch as creating a new market category.

Viewed through a barbell approach, the lesson is that an operating business can preserve capacity for ideas whose applications and returns are initially uncertain. The connection to Blue Ocean Strategy is the search for a different source of customer value. The example alone does not demonstrate every element of that framework, including its low-cost requirement.

**How this Applies to You**

You should make room for two different kinds of product work and evaluate them appropriately. An established product should demonstrate dependable performance. An early experiment should establish whether a meaningful opportunity exists. Requiring immediate revenue from every experiment can prevent useful learning, while allowing exploration to continue without evidence can waste substantial resources.

Make the limits of each experiment concrete. Define the question, budget, duration, and evidence needed for the next investment. Consider exposure beyond spending: a small pilot can still damage customer trust or disrupt an essential service. Also question whether the core is truly dependable. An established product facing declining demand cannot automatically serve as the secure side of the portfolio.

When exploring new demand, spend time with people who avoid or cannot use the existing category. Find out what feels too expensive, complicated, inconvenient, or unnecessary. Then test whether removing those barriers creates an offering they value. Your responsibility is to protect the business’s ability to keep operating while giving well-supported opportunities enough room to develop.

**Continue Down the Rabbit Hole …**

* **Innovation Doesn’t Have to Be Disruptive** [*Harvard Business Review*]
  [https://store.hbr.org/product/innovation-doesn-t-have-to-be-disruptive/R2303D](https://store.hbr.org/product/innovation-doesn-t-have-to-be-disruptive/R2303D)

* **Listen to users, but only 85% of the time** [*UX Collective*]
  [https://uxdesign.cc/listen-to-users-but-only-85-of-the-time-da96c2f9dde6](https://uxdesign.cc/listen-to-users-but-only-85-of-the-time-da96c2f9dde6)

* **Innovation Systems Need a Reboot** [*BCG*]
  [https://bcg.com/publications/2024/innovation-systems-need-a-reboot](https://bcg.com/publications/2024/innovation-systems-need-a-reboot)

* **Introducing the Light Generation** [*Google X*]
  [https://x.company/blog/posts/taara-graduation/](https://x.company/blog/posts/taara-graduation/)

# Pareto Principle [80/20 Rule]

Introduction: The Pareto Principle, commonly called the 80/20 rule, describes a pattern where a relatively small share of inputs accounts for a large share of results. In product development, a few problems may generate most customer complaints, or a small number of products may produce most sales. The exact split does not need to be 80/20. The practical lesson is to investigate where results are concentrated before deciding where to spend your time as a team’s capacity is limited, and some improvements will matter much more than others.

**The Need to Know**

A product backlog can make every request look equally deserving of attention. Each item occupies a row, has an owner, and comes with a reason to build it. But fixing ten minor annoyances may accomplish less than resolving one problem that prevents customers from completing their work. Pareto encourages designers and managers to look for that concentration of impact.

To apply it, choose a specific outcome and measurement period. You might examine product returns over the last quarter, support hours spent on different issues, or revenue by product line. Rank the categories from largest to smallest and calculate their cumulative share of the total. A Pareto chart makes this pattern visible through descending bars and, often, a cumulative percentage line. The American Society for Quality recommends this approach when teams need to identify the most significant problems among many competing candidates.

The measurement matters. The most frequent complaint may be inexpensive to resolve, while a less frequent failure creates substantial customer harm or cost. Similarly, the most-used feature is not automatically the most valuable. Customers rarely use account recovery until they urgently need it. Pareto helps you identify where to investigate; you still need judgment about severity, customer needs, and the effort required to make an improvement.
**The Real-world Application**

**LEGO’s** turnaround in the mid-2000s offers a concrete example of why concentration matters. A strategy+business account reported that just 30 products generated 80% of LEGO’s sales, even as its assortment and operations had become increasingly complex. Designers were adding colors, components, and specialized materials without consistently accounting for the resulting costs. A relatively small group of products drove most sales, while the broader system carried substantial complexity.

As part of its response, LEGO made the costs of new colors and shapes visible to designers and encouraged creative reuse of existing elements. The article reports that component cost analysis, combined with a sourcing initiative, helped cut resin costs in half. This illustrates Pareto-style thinking within a broader turnaround: identify where value and cost are concentrated, then focus the work accordingly. It does not establish an exact 80/20 split across every part of the business.

For a product team, the question is whether each additional variation earns its development and maintenance cost. More options can benefit customers, but they also create work in manufacturing, testing, support, and inventory. Before expanding the product, understand which options customers value and which mainly add complexity.

**How this Applies to You**

You should become more deliberate about where you spend the team’s attention. Ask which customer problems account for the largest share of lost time, failed tasks, or dissatisfaction. Then investigate why those problems occur. A category that dominates support tickets identifies a promising area for research, but it does not establish the root cause or tell you which solution will work.

You also need to protect against treating the remaining customers or requirements as disposable. A rarely used accessibility feature may be essential to the people who depend on it. An emerging product may have low sales because it is new. Historical concentration should inform your priorities while leaving room for essential requirements and future opportunities.

**Continue Down the Rabbit Hole …**

* **What Is the Pareto Principle?** [*Interaction Design Foundation*]
  [https://ixdf.org/literature/topics/pareto-principle](https://ixdf.org/literature/topics/pareto-principle)

* **Pareto Principle: How the 80/20 Rule Helps Teams Prioritize Work** [*Atlassian*]
  [https://atlassian.com/work-management/strategic-planning/pareto-principle](https://atlassian.com/work-management/strategic-planning/pareto-principle)

* **Pareto Principle 80/20 Rule: Prioritize for Team** [*Asana*]
  [https://asana.com/resources/pareto-principle-80-20-rule](https://asana.com/resources/pareto-principle-80-20-rule)

* **The 80/20 Rule for Agile Product Managers** [*ProductPlan*]
  [https://productplan.com/learn/80-20-rule-agile](https://productplan.com/learn/80-20-rule-agile)

# Data-Driven Decision Making

Introduction: Every product team has opinions about what customers want and what the business should build next. Data-driven decision making gives us a way to examine those opinions using evidence. It means defining the decision, choosing measures that reflect the outcome we care about, and using what we learn to guide our actions. In Predictable Product Development, this connects customer behavior and business results to product choices. The goal is to make better decisions with a clear understanding of what the evidence supports, what remains uncertain, and what would change our minds.

**The Need to Know**

A number can be accurate and still tell you very little about whether a product is succeeding. Downloads show that people tried something. They don't establish that anyone found it useful. Time spent in a product might reflect engagement, or it might mean customers are struggling to finish a task. Product designers and managers need to understand the relationship between the metric and the outcome before treating an increase as progress.

Data also needs context. Imagine that customer retention improves after a redesign. The redesign might have helped, but perhaps the company also changed its pricing or attracted a different type of customer. Correlation alone doesn't establish causation. A properly designed randomized experiment can help isolate the effect of a change. When experimentation isn't practical, teams should compare relevant groups and time periods while being explicit about other possible explanations.

Quantitative and qualitative evidence work together. Usage data can reveal where customers abandon a process. Interviews and observation can help explain what they found confusing or unnecessary. You also need to know whose experience the data represents. Looking only at active customers can hide the problems that drove other people away. A useful analysis includes the people who struggled, stopped, or never completed the first step.

**The Real-world Application**

**Duolingo** and its team noticed that some learners practiced every day but still failed to maintain a streak because they hadn't completed their full daily goal. People with more ambitious goals were particularly affected. The team tested separating the two mechanics: completing one lesson would extend a streak, while progress toward the daily goal would be tracked separately. The A/B test produced a 3.3% relative increase in retention after 14 days and a 1% increase in daily active learners.

The results also exposed a tradeoff. Fewer learners reached their daily goals, but more kept returning, and total time spent learning increased. Duolingo chose to prioritize consistency. This illustrates why product teams need several measures to interpret an outcome: a decline in daily goal completion could look like failure in isolation. At the same time, more activity doesn't establish greater language proficiency, which would require separate evidence.

**How this Applies to You**

It is imperative that product leaders be able to explain why a metric matters. A dashboard full of activity measures can create the appearance of progress without helping anyone decide what to do. Connect each important measure to a customer or business outcome, and ask what action you would take if it increased, decreased, or stayed the same. If none of those results would affect a decision, reconsider why you are tracking it.

You also need to become more willing to revise a preferred idea. Decide what evidence would change your mind before the results arrive. Examine findings that contradict your expectations with the same care you give findings that support them. At the same time, recognize the limits of historical data: it describes what happened under previous conditions and may say little about a genuinely new opportunity. Research, prototypes, and small experiments can help you learn where existing data falls short.

Before presenting a recommendation, ask: What decision does this evidence support, and what else could explain the result? Your responsibility is to turn the answer into a clear course of action. That may mean launching a change, revising it, stopping work, or gathering more evidence because the remaining uncertainty could materially affect the decision.

**Continue Down the Rabbit Hole …**

* **Aligning UX Metrics with Organizational Goals** [*Nielsen*]
  [https://nngroup.com/articles/ux-metrics-goals/](https://nngroup.com/articles/ux-metrics-goals/)

* **The Surprising Power of Online Experiments** [*Harvard Business Review*]
  [https://hbr.org/2017/09/the-surprising-power-of-online-experiments](https://hbr.org/2017/09/the-surprising-power-of-online-experiments)

* **Better Experiments with LLM Evals: A Funnel, Not a Fork** [*Spotify*]
  [https://engineering.atspotify.com/2026/5/better-experiments-with-llm-evals-a-funnel-not-a-fork](https://engineering.atspotify.com/2026/5/better-experiments-with-llm-evals-a-funnel-not-a-fork)

* **Don't Let Metrics Undermine Your Business** [*Harvard Business Review*]
  [https://hbr.org/2019/09/dont-let-metrics-undermine-your-business](https://hbr.org/2019/09/dont-let-metrics-undermine-your-business)

# Continuous Delivery Fundamentals

Introduction: Continuous delivery is the ability to keep software ready for a safe, reliable release whenever the team decides it should reach customers. It depends on small changes, frequent integration, automated checks, and a repeatable deployment process. This is a valuable connection between building and learning: teams can deliver a useful improvement, observe its effect, and adjust without waiting for a major launch. Continuous delivery keeps releases available on demand. Continuous deployment goes further by automatically deploying changes that pass the required checks.

**The Need to Know**

Large releases bundle many decisions together. If the customer experience improves, it can be difficult to identify which change helped. If something breaks, the team has more potential causes to investigate. Smaller changes can make problems easier to isolate and reduce the amount of unfinished work waiting to reach customers. They also give teams more opportunities to learn before committing to an entire feature or redesign.

The foundation is a dependable path from a proposed change to working software. Developers combine their work frequently, a practice called continuous integration. Automated processes build and test the software, while additional checks assess whether it is ready to release. A repeatable deployment process moves it into production, where monitoring helps the team detect problems. Quality checks happen throughout development, with human review and exploratory testing where needed.

For designers and product managers, this creates more flexibility in how work is divided. A team improving account setup might first simplify one confusing step, then evaluate the result before changing the rest. Each release should still provide a coherent experience. Delivering smaller pieces requires careful product judgment about what customers can use meaningfully and what should remain unavailable until it is ready.

**The Real-world Application**

Waymo illustrates how continuous delivery principles apply to a product where software changes affect physical safety. Its development process combines simulation, closed-course testing, and driving on public roads. In simulation, teams can replay situations encountered by vehicles and introduce variations to evaluate how software responds; creating a repeatable way to assess improvements before exposing passengers and other road users to them.

Waymo also separates development progress from deployment readiness. Its published safety approach describes criteria used to assess readiness before deployment, followed by analysis of operational performance to check whether the expected outcomes occur. This illustrates a central continuous delivery principle: teams need a dependable process for evaluating and releasing changes, with evidence supporting the decision to deploy. It does not mean every software change automatically reaches the fleet.

We use continuous delivery fundamentals to define release conditions alongside the intended improvement. Suppose a team is improving how an autonomous vehicle handles an unfamiliar road layout. It needs to demonstrate that the change addresses the problem, check that existing capabilities still work, and monitor performance after deployment. The practical question becomes: “What evidence would make us confident this improvement is ready for customers?” Continuous delivery makes answering that question a repeatable part of development.

**How this Applies to You**

You should think about the smallest complete improvement that can reach a customer and produce useful evidence. Designers need to ensure that intermediate releases make sense and fit the broader experience. Product owners need to define the intended outcome, select an appropriate audience, and decide what would justify expanding, revising, or stopping the rollout. These decisions should be made with engineering before the work is considered finished.

You also need to help make delivery dependable. Allow time for testing, deployment automation, monitoring, and recovery planning. Those capabilities affect how confidently the team can respond to customers later. For physical products, frequent software updates may be possible, while tooling changes and manufactured components require different release schedules. Apply the principle of small, validated changes with respect for those constraints.

**Continue Down the Rabbit Hole …**

* **Continuous Integration** [*Martin Fowler*]
  [https://martinfowler.com/articles/continuousIntegration.html](https://martinfowler.com/articles/continuousIntegration.html)

* **The Evolution of Software Delivery Performance** [*Continuous Delivery Foundation*]
  [https://cd.foundation/state-of-cicd-2024/](https://cd.foundation/state-of-cicd-2024/)

* **Accelerate State of DevOps Report** [*Google Cloud*]
  [https://dora.dev/research/2024/dora-report/](https://dora.dev/research/2024/dora-report/)

* **Accelerating Large-Scale Test Migration with LLMs** [*Airbnb*]
  [https://medium.com/airbnb-engineering/accelerating-large-scale-test-migration-with-llms-9565c208023b](https://medium.com/airbnb-engineering/accelerating-large-scale-test-migration-with-llms-9565c208023b)

#

# Real-Time Feedback Loops

Introduction: Real-time feedback loops connect what is happening in a product or process to a timely response, then measure whether that response helped. The basic cycle is observe, interpret, act, and measure again. This means designing how you will learn alongside what you will build; deciding which signals matter, who will respond, and what evidence will show improvement. “Real-time” means fast enough to influence the situation while it still matters.

**The Need to Know**

The longer a team takes to discover a problem, the more customers may experience it and the more expensive it can become to correct. A confusing setup step can drive abandonment for weeks before a scheduled research session reveals it. A manufacturing defect can spread across thousands of units before returns expose the pattern. Timely feedback gives teams a chance to intervene while the problem is still contained.

Feedback can come from customer behavior, direct comments, support interactions, equipment sensors, or product performance data. Each tells you something different. A sudden increase in failed checkout attempts identifies a problem, but customer conversations or technical investigation may be needed to explain it. Useful loops connect detection to diagnosis and give someone the authority to respond. Collecting more information without establishing that responsibility simply creates more information to review.

Speed also requires judgment. A drop in task completion immediately after a release may justify investigation or a rollback. A small fluctuation in daily engagement may reflect ordinary variation. Teams should define thresholds, escalation rules, and success measures before they face pressure to react. Live data can reveal problems quickly, while conclusions about retention, durability, or lasting customer value may still require weeks or months of observation.

**The Real-world Application**

**Toyota’s** Andon system is a clear example of a feedback loop built into everyday work. When a worker identifies a production problem, a cord or button can signal where assistance is needed. The team leader responds, and the line can stop if the issue cannot be resolved within the available time. This connects a local observation to an immediate operational response.

The response also feeds longer-term improvement. Toyota describes analyzing the frequency and nature of these problems as part of its continuous improvement process. Its broader production philosophy, *Jidoka*, includes detecting abnormalities, stopping when necessary, and preventing recurrence. The loop therefore operates at two speeds: address the immediate issue, then improve the process that allowed it to occur.

A product team can apply the same logic to a new release. Suppose customers suddenly begin failing an account setup step. Monitoring flags the change, a responsible team investigates, and the team fixes or reverses the release. It then checks whether successful setup has recovered and updates its testing process. The transferable lesson is that the signal, response, and follow-up need to be designed together.

**How this Applies to You**

You should consider how a product will reveal its own problems. Designers can make errors understandable, provide recovery paths, and give customers a useful way to report difficulty. Product leaders can define the outcomes to monitor and establish who acts when performance changes. Engineering and operations partners help make those signals dependable and the responses practical.

Separate immediate correction from evidence that a design is better. Restoring service after an error is one kind of success. Demonstrating that a redesigned workflow improves customer outcomes requires an appropriate comparison, sufficient observation, and attention to unintended effects. A change that increases completion while also increasing mistakes may need further work. The loop should measure the customer’s result, not just the action your team wants them to take. The important thing to know is, “If this experience failed tomorrow, how quickly would we know, what would we do, and how would we know we had fixed it?”

**Continue Down the Rabbit Hole …**

* **The Signals Loop: Fine-tuning for World-Class AI Apps and Agents** [*Microsoft* *Azure*]
  [https://azure.microsoft.com/blog/the-signals-loop-fine-tuning-for-world-class-ai-apps-and-agents/](https://azure.microsoft.com/blog/the-signals-loop-fine-tuning-for-world-class-ai-apps-and-agents/)

* **Beyond Winning: Spotify’s Experiments with Learning Framework** [*Spotify*]
  [https://engineering.atspotify.com/2025/9/spotifys-experiments-with-learning-framework](https://engineering.atspotify.com/2025/9/spotifys-experiments-with-learning-framework)

* **Sequential A/B Testing Keeps the World Streaming Netflix** [*Netflix*]
  [https://netflixtechblog.com/sequential-a-b-testing-keeps-the-world-streaming-netflix-part-1-continuous-data-cba6c7ed49df](https://netflixtechblog.com/sequential-a-b-testing-keeps-the-world-streaming-netflix-part-1-continuous-data-cba6c7ed49df)

#

# The Data Feedback Loop

Introduction: The Data Feedback Loop is the process through which product use generates information, that information informs a change, and the results of that change generate new evidence. It connects observation, interpretation, action, and measurement into a repeating cycle. For product designers and managers, the value is learning whether the product actually delivers the intended benefit and using that knowledge to improve it. A loop can operate automatically within a recommendation system or through a team reviewing research and making design changes.

**The Need to Know**

Launching a product tests assumptions about customer needs, behavior, and willingness to pay. Some will hold up; others will need revision. A Data Feedback Loop gives teams a repeatable way to investigate those assumptions after the product reaches users. Without it, a team can keep adding functionality while the same underlying problems persist.

| Stage | What happens | Example(s) |
| :---- | :---- | :---- |
| **Observe** | Collect relevant behavioral data and customer feedback | Track completion, errors, and requests for help |
| **Interpret** | Investigate patterns and competing explanations | Determine whether abandonment reflects confusion, technical failure, or low interest |
| **Act** | Make a targeted change based on a clear hypothesis | Simplify a step that customers repeatedly misunderstand |
| **Measure** | Evaluate the effect against the intended outcome | Check whether more customers complete setup successfully without creating additional errors |
| **Repeat** | Use the findings to guide the next decision | Expand the change, revise it, or investigate another explanation |

The loop’s quality depends on what you measure. More clicks might indicate interest, but they might also indicate difficulty finding something. Fewer support requests could mean an easier experience or a harder-to-find help button. Pair behavioral metrics with customer research and measures of unintended consequences. When feasible, use a controlled experiment to distinguish the effect of a change from seasonality, audience shifts, or other factors.

Feedback can also reinforce mistakes. A recommendation system that repeatedly shows the same category will collect more interactions with that category, potentially mistaking exposure for preference. More data does not automatically correct this problem. Teams need ways to test alternatives, examine who is missing from the evidence, and recognize when the product itself shapes the behavior being measured.

**The Real-world Application**

**Waze** makes the Data Feedback Loop visible in everyday use. Drivers using the app contribute information about road conditions and travel speeds. They can also submit reports about traffic, accidents, and blocked roads. Waze explains that it analyzes these inputs to inform navigation and compares information from roads driven with the app open when evaluating routes.

The loop connects usage with utility: driving generates observations, those observations inform route guidance, and subsequent journeys generate additional observations. A traffic estimate therefore has a way to change as conditions change. The customer benefit comes from turning fresh information into a useful decision about where to drive.

For product teams, the broader lesson is to design the connection between collecting information and acting on it. A hypothetical navigation change should be evaluated using actual journey outcomes, prediction errors, and relevant side effects. A route being selected does not by itself establish that it was better. The team still needs evidence that the guidance improved the experience.

**How this Applies to You**

Design the learning process alongside the feature. Before release, define the expected outcome, the information needed to evaluate it, and who will review the results. Decide what evidence would justify expanding, revising, or reversing the change. This makes measurement part of development rather than something added after questions arise.

Stay close to interpretation. Designers can help explain observed behavior through interviews and usability testing. Product managers can connect those findings to customer value, business performance, and priorities. Engineers and analysts can verify that instrumentation and comparisons are reliable. A dashboard becomes useful when someone has the responsibility and authority to act on what it reveals.

Choose a feedback cadence that matches the decision. Some conditions require an immediate response; other outcomes need weeks or months to evaluate meaningfully. Allow enough time for the intended effect to appear, and revisit assumptions when the audience or environment changes.

**Continue Down the Rabbit Hole …**

* **Risk-Aware Product Decisions in A/B Tests with Multiple Metrics** [*Spotify*]
  [https://engineering.atspotify.com/2024/3/risk-aware-product-decisions-in-a-b-tests-with-multiple-metrics](https://engineering.atspotify.com/2024/3/risk-aware-product-decisions-in-a-b-tests-with-multiple-metrics)

* **Experimentation in Generative AI: Practices for Continuous Improvement** [*Microsoft*]
  [https://microsoft.com/en-us/research/articles/experimentation-in-genai-c-teams-practices-for-continuous-improvement/](https://microsoft.com/en-us/research/articles/experimentation-in-genai-c-teams-practices-for-continuous-improvement/)

* **How Human–AI Feedback Loops Alter Human Perceptual, Emotional and Social Judgements** [*Nature Human Behaviour*]
  [https://nature.com/articles/s41562-024-02077-2/](https://nature.com/articles/s41562-024-02077-2/)

# Data Radar [Framework]

Introduction: The Data Radar framework helps product teams build a broader view of the evidence informing their decisions. It brings together four sources: behavioral data, attitudinal data, market and external data, and operational and financial data. Together, they reveal what customers do, what they say and need, what is changing around the business, and whether the product can deliver sustainable value. The framework then connects that evidence to the business mission, customer benefits, competitive advantage, and the resources needed to act. Its purpose is to make sure consequential product decisions account for the signals that matter, including those outside the team’s usual field of view.

**The Need to Know**

Every data source provides an incomplete picture. Usage analytics can show where customers abandon a task, but may not explain their frustration. Interviews can reveal a strong need without establishing how widespread it is. Market research can identify an opportunity without showing whether your business can serve it profitably. A Data Radar encourages teams to bring these perspectives together before committing to a direction.

| Data source | What it contributes | Examples |
| :---- | :---- | :---- |
| **Behavioral Data** | Evidence of what people actually do | Clickstreams, feature usage, purchase behavior, and conversion funnels |
| **Attitudinal Data** | Insight into needs, motivations, expectations, and perceptions | Interviews, surveys, customer feedback, and jobs-to-be-done research |
| **Market & External Data** | Context about alternatives and changes outside the business | Competitor benchmarks, industry trends, economic indicators, and public datasets |
| **Operational & Financial Data** | Evidence of delivery constraints and business performance | Service costs, revenue, retention, customer acquisition cost, and collection times |

The most useful insights often emerge when sources disagree. Customers might praise a feature in interviews while rarely using it. Adoption might increase while support costs make the feature expensive to maintain. These differences deserve investigation. They may expose a gap between stated preferences and actual needs, a problem in measurement, or a tradeoff the team has overlooked.

To make the radar actionable, start with the business mission and define the role of data in supporting it. Identify the unique value of data, the benefit it creates for customers, and the desired future state. Then specify data governance and the people and resources required. For each element, record short-term goals, long-term goals, required resources, and success metrics. This turns a broad scan of information into a practical plan.

**The Real-world Application**

**Stitch Fix** illustrates how several kinds of evidence can inform one product experience. Its published Algorithms Tour describes combining customer style profiles, written feedback, purchase histories, and clothing attributes to support recommendations. It also describes using data to manage inventory, assign warehouses, forecast demand, and support human stylists; connecting customer preferences with the operational work required to fulfill them.

Viewed through Data Radar, purchase histories provide behavioral evidence, while stated preferences and fit feedback provide attitudinal evidence. Inventory availability, shipping considerations, and carrying costs contribute to operational and financial context. The company also describes sourcing merchandise from the market and developing styles for underserved customer segments.

For a team building a similar service, the planning step would connect those inputs to a clear customer promise: finding suitable clothing with less effort. An initial goal might be to improve fit recommendations; a longer-term goal might be to anticipate demand more accurately. The plan would identify the necessary research, inventory information, technical capabilities, and rules for handling customer data. Success would need to include both customer outcomes and the cost of delivering them.

**How this Applies to You**

Take responsibility for the range of evidence behind a decision. Before approving an initiative, ask whether the team understands customer behavior, customer motivations, market conditions, and business implications. The balance will vary by decision. A usability fix may need observation and interviews most urgently, while entering a new market requires substantial external and financial evidence.

Give each data activity a purpose. Collecting information, commissioning research, or building a dashboard should help resolve a question or improve an experience. Decide who owns the work, what resources it requires, and how the findings will influence a choice. Address access, privacy, quality, and appropriate use when planning the work.

Keep the radar active as conditions change. Revisit assumptions when customer behavior shifts, competitors introduce alternatives, or delivery costs change. When signals conflict, define the next investigation instead of choosing the source that supports your preferred answer. The question to keep asking is: “What important signal are we missing, and how could it change our decision?”

**Continue Down the Rabbit Hole …**

* **Data and AI Ethics Framework** [*UK Government Digital Service*]
  [https://gov.uk/government/publications/data-ethics-framework](https://gov.uk/government/publications/data-ethics-framework)

* **Artificial Intelligence Risk Management Framework** [*NIST*]
  [https://nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence](https://nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence)

* **Disempowerment Patterns in Real-World AI Usage** [*Anthropic*]
  [https://anthropic.com/research/disempowerment-patterns](https://anthropic.com/research/disempowerment-patterns)

* **Recruiting and Screening Candidates for User Research Projects** [*Nielsen*]
  [https://nngroup.com/articles/recruiting-screening-research-candidates/](https://nngroup.com/articles/recruiting-screening-research-candidates/)

Guide to Radar Charts: Concepts, Examples, Templates by *ProcessOn* A ProcessOn guide with editable radar chart templates for comparing products, skills, or performance across multiple dimensions.[https://processon.io/blog/radar-chart-template/](https://processon.io/blog/radar-chart-template/)

# Dimensionalities of Data

Introduction: Dimensionalities of data refers to the different variables used to describe and analyze data through five dimensions: quality, bias, discrimination, personalization, and diversity. Together, they help us evaluate whether the information behind a product is trustworthy, whom it represents, and how its use affects customers. The opportunity is to look beyond whether a team has enough data to make a decision. You also need to ask whether that data supports the decision you are making and whether the resulting experience works across the people you intend to serve. These five dimensions provide a practical lens for evaluating data throughout product development, from initial customer research to personalized recommendations after launch.

**The Need to Know**

Data influences whose problems get prioritized, what a product recommends, and how success is measured. A large dataset can still contain inaccurate information or exclude important customer groups. A technically accurate prediction can also produce an unfair experience when used to make the wrong kind of decision. These five dimensions help product teams examine both the evidence they use and the consequences of using it.

| Dimension | Description |
| :---- | :---- |
| **Data Quality** | Ensuring data is accurate, complete, relevant, and timely enough to support reliable product decisions |
| **Data Bias** | Recognizing how skewed data collection, sampling, or interpretation can distort understanding of users and lead to flawed product decisions |
| **Data Discrimination** | Examining how data-driven product decisions can unfairly exclude, disadvantage, or treat certain users differently |
| **Data Personalization** | Using data about users’ needs, behaviors, and preferences to tailor product experiences while respecting privacy and user control |
| **Data Diversity** | Incorporating data from varied users, contexts, and sources to uncover overlooked needs and build products that serve a broader range of people |

These dimensions are connected, but each reveals a different problem. Survey responses might be recorded perfectly, giving you accurate records, while the survey reaches only your most enthusiastic customers, creating selection bias. Using those findings to remove support options could then disadvantage customers whose needs were never represented. Adding more responses from the same narrow audience would not resolve that gap.
Personalization brings these issues into the product experience. A system can learn from previous behavior and become more useful, but it can also reinforce an incorrect assumption. Someone buying a gift should not necessarily have their future recommendations built around the recipient’s preferences. Good personalization gives users ways to express intent, correct mistakes, and explore beyond what the system already knows.

**The Real-world Application**

L’Oréal’s Beauty Tech services offer a useful example. The company describes virtual tools for trying makeup and hair colors, along with AI-supported skin and hair analysis that informs personalized routines and product recommendations. These services connect information about the individual with a product choice, making data central to the customer experience.

Applying our five dimensions to a service like this creates concrete product questions. Quality means checking whether images and customer inputs are suitable for the task. Bias means investigating whether the system performs better under certain lighting conditions or for certain users. Diversity means testing across relevant skin tones, hair textures, ages, devices, and environments. Personalization means making recommendations responsive to the person’s stated needs and preferences. Discrimination means examining whether some groups consistently receive a less useful experience or fewer suitable options.

Those are evaluation questions, not claims about problems in L’Oréal’s systems or proof that it has resolved them. For a team developing a similar service, overall recommendation accuracy would be only one measure. The team should also investigate performance across customer groups, whether people can correct inaccurate inputs, and whether recommendations help them make a satisfactory choice.

**How this Applies to You**

Treat decisions about data as part of designing the product. Define what information is necessary, where it comes from, and what it can reasonably tell you. Customer research, analytics, and model outputs all involve choices about whom to include and what to measure. Ask for those choices to be made explicit before treating the results as reliable evidence.

Distinguish useful adaptation from unfair exclusion. Different customers may need different experiences, but those differences should have a defensible connection to their needs. Give users appropriate control over personalization and investigate meaningful performance gaps. Removing a demographic field alone does not guarantee fair outcomes because other variables can carry related information.

**Continue Down the Rabbit Hole …**

* **Data Quality: What Is Data Quality?** [*IBM*]
  [https://ibm.com/think/topics/data-quality/](https://ibm.com/think/topics/data-quality/)

* **Data Bias: Everyone Wants to Do the Model Work, Not the Data Work** *[Google Research]*
  [https://research.google/pubs/everyone-wants-to-do-the-model-work-not-the-data-work-data-cascades-in-high-stakes-ai/](https://research.google/pubs/everyone-wants-to-do-the-model-work-not-the-data-work-data-cascades-in-high-stakes-ai/)

* **Data Discrimination: Rite Aid Banned from Using AI Facial Recognition After FTC Says Retailer Deployed Technology without Reasonable Safeguards** [*FTC*]
  [https://ftc.gov/news-events/news/press-releases/2023/12/rite-aid-banned-using-ai-facial-recognition-after-ftc-says-retailer-deployed-technology-without](https://ftc.gov/news-events/news/press-releases/2023/12/rite-aid-banned-using-ai-facial-recognition-after-ftc-says-retailer-deployed-technology-without)

* **Data Personalization: Spotify Premium Users Can Now Turn Any Idea Into a Personalized Playlist With AI Playlist in Beta** [*Spotify*]
  [https://newsroom.spotify.com/2024-04-07/spotify-premium-users-can-now-turn-any-idea-into-a-personalized-playlist-with-ai-playlist-in-beta/](https://newsroom.spotify.com/2024-04-07/spotify-premium-users-can-now-turn-any-idea-into-a-personalized-playlist-with-ai-playlist-in-beta/)

* **Data Diversity: Bridging the Data Provenance Gap Across Text, Speech and Video** [*ICLR Research*]
  [https://arxiv.org/abs/2412.17847](https://arxiv.org/abs/2412.17847)

#

# Data Co-opetition

Introduction: Data co-opetition combines cooperation and competition: companies collaborate on selected data, standards, or infrastructure while continuing to compete for customers. The premise is that some problems become easier to solve when businesses contribute information that no single company can collect alone. For product designers and managers, this creates a strategic choice about what to share and where to differentiate. Shared information might improve reliability, traceability, or fraud detection, while each company competes through its product experience, service, and execution.

**The Need to Know**

A company’s own data provides an incomplete view of the world. A manufacturer may know how a component performs in its products but have limited visibility into problems elsewhere in the supply chain. Competitors can face the same information gaps and spend considerable resources solving them separately. Selective collaboration can improve coverage and reduce duplicated work.

The strategic question is which information creates more value when shared. Some data supports capabilities that an entire industry needs, such as identifying defective components or interpreting material specifications consistently. Other information supports a company’s distinctive advantage. Teams need to make that distinction deliberately rather than assuming that every dataset should be either fully private or openly available.

| Decision | Question for the product team |
| :---- | :---- |
| **Shared purpose** | What customer or operational problem can participants solve better together? |
| **Contribution** | What useful data will each participant provide, and how will quality be checked? |
| **Access and use** | Who can access the information, for which purposes, and for how long? |
| **Competitive boundary** | Which information and product decisions remain independent? |
| **Mutual benefit** | Why will participants continue contributing rather than only consuming? |
| **Success** | What measurable improvement would justify the cost of collaboration? |

**The Real-world Application**

BMW Group participates in Catena-X, an automotive data ecosystem that also includes competing manufacturers such as Mercedes-Benz and Volkswagen. Participants work toward common standards for exchanging information across the industry’s supply chains. This illustrates co-opetition at the infrastructure level: competitors support a shared way to exchange data while retaining their separate businesses and product strategies.

BMW describes applications involving material composition, carbon footprints, and notifications about defective components. For example, a supplier can communicate that a component should be blocked from use, helping the receiving manufacturer respond earlier. Catena-X combines standardized formats with rules governing data exchange and participants’ control over their information. Participation does not mean every member can freely inspect every other member’s records.

For a product team, the useful distinction is between a common information foundation and the decisions made with it. Manufacturers can benefit from compatible component data while making different choices about engineering, sourcing, service, and customer experience. The collaboration reduces friction in exchanging information. Each participant still has to turn that information into better products and operations.

**How this Applies to You**

Expand your thinking beyond the data your organization already holds. Start with the decision you need to improve, then identify who has relevant information and why they might participate. A useful proposal should explain the benefit to the partner as clearly as the benefit to your own business. Access without a reason to keep contributing is unlikely to produce a dependable partnership.

Treat the exchange itself as a product experience. Partners need understandable definitions, manageable onboarding, clear permissions, and ways to correct errors. Your internal users need to know where information came from, how current it is, and what limitations affect its use. These details determine whether shared data becomes useful in everyday work.

Finally, be explicit about where your product will continue to earn customer preference. If several competitors can access similar information, your advantage may come from interpretation, workflow design, reliability, or service. Measure the collaboration through outcomes such as faster issue detection, fewer defects, or reduced integration work. The amount of data exchanged matters only when it improves a decision or experience.

**Continue Down the Rabbit Hole …**

* **EU Code of Conduct for Data Sharing in the Social Economy** [*European Commission*]
  [https://social-economy-gateway.ec.europa.eu/eu-initiatives/code-conduct-data-sharing-social-economy_en](https://social-economy-gateway.ec.europa.eu/eu-initiatives/code-conduct-data-sharing-social-economy_en)

* **Powering Microsoft Maps with Overture: Faster Releases, Better Data** [*Overture*]
  [https://overturemaps.org/case-study/2026/powering-microsoft-maps-with-overture-faster-releases-better-data/](https://overturemaps.org/case-study/2026/powering-microsoft-maps-with-overture-faster-releases-better-data/)

* **Meta Expands Data Accord with Banks to Cut Online Fraud** [*Financial Times*]
  [https://ft.com/content/ff5ed0ed-170d-4439-a789-d658936fd5f5](https://ft.com/content/ff5ed0ed-170d-4439-a789-d658936fd5f5)

# Metcalfe's Law

Introduction: Metcalfe’s Law proposes that a network’s value grows roughly in proportion to the square of its number of connected participants. The idea is that each new participant creates opportunities for connections with everyone already there. For product designers and managers, this helps explain why a product can become more useful as more people join. A messaging tool, professional network, or collaboration platform depends partly on who else uses it. However, the formula describes potential value. Actual value depends on whether people make useful connections, participate, and trust the experience.

**The Need to Know**

Some products gain value through the relationships they enable. A collaboration tool becomes more useful when your teammates join because you can coordinate work in one place. This is a network effect: additional participants improve the experience for existing participants. It is different from viral growth, which describes how a product spreads. Invitations can drive signups without creating lasting value if the people who join never interact.

The math illustrates the opportunity. In a network where every participant could connect with every other participant, the number of unique pairs is n(n − 1) / 2.

Doubling participation can more than quadruple the number of possible connections, but those connections are not equally valuable. A thousand irrelevant contacts may contribute less than five people you regularly work with. Spam, inactivity, and overcrowding can also make a growing network less useful. This limitation is central to published critiques of Metcalfe’s Law: counting connections does not establish their economic value. Product teams should use the law to examine opportunities for interaction, while testing whether growth actually improves the experience.

**The Real-world Application**

**LinkedIn** illustrates how a product can create value by helping people build relevant connections. Its “People You May Know” feature suggests potential contacts based on shared connections, profile information, professional experience, and other commonalities. These recommendations help members find people within a network that would be difficult to navigate manually. LinkedIn describes relevance and scale as two central challenges in building this feature.

Viewed through Metcalfe’s Law, each additional professional creates potential connections with existing members. A new member might become someone’s future colleague, mentor, client, or employer. The product’s job is to help people discover which of those connections matter. A larger membership creates more possibilities, while recommendations, profiles, and communication tools help turn those possibilities into useful interactions. This illustrates the principle without establishing that LinkedIn’s business value follows an exact squared relationship.

For a team building a professional community, the practical implication would be to start with a group whose members have clear reasons to interact. A focused community of industrial designers might initially create value through portfolio feedback, specialist advice, and project referrals. The team could measure how quickly new members have a useful exchange, how often conversations receive responses, and whether participants return. These measures would help establish whether the network is becoming more valuable as it grows.

**How this Applies to You**

Metcalfe’s Law should encourage you to design for relationships between users. Alongside asking what someone can accomplish individually, ask what becomes possible when the right people participate together. That changes decisions about onboarding, invitations, discovery, permissions, and collaboration. Helping someone reach their first useful interaction may matter more than adding another feature.

It also changes how you approach early growth. A new network faces a practical problem: people may hesitate to join until others are already participating. You need a credible reason for the first group to show up and stay. That could mean serving one tightly connected community, supporting an existing team, or providing useful individual functionality while the network develops.

Finally, measure whether participation improves outcomes. Look at active relationships, response rates, successful collaborations, retention, and unwanted interactions alongside membership growth. Build the controls that preserve trust as participation expands. Your responsibility is to make it easier for people to find and sustain connections that are worth their time.

**Continue Down the Rabbit Hole …**

* **Network Effects: Definition, Types, and Examples** [*Shopify*]
  [https://shopify.com/blog/network-effects](https://shopify.com/blog/network-effects)

* **Expanding Stripe Radar to Protect More of Your Business** [*Stripe*]
  [https://stripe.com/blog/expanding-stripe-radar-to-protect-more-of-your-business/](https://stripe.com/blog/expanding-stripe-radar-to-protect-more-of-your-business/)

* **Marketplaces in the Age of AI, Take Two: Graveyard to Greenfield** [*a16z*]
  [https://a16z.com/marketplaces-in-the-age-of-ai-take-two-graveyard-to-greenfield/](https://a16z.com/marketplaces-in-the-age-of-ai-take-two-graveyard-to-greenfield/)

* **Metcalfe’s Law Is Wrong** [*IEEE Spectrum*]
  [https://spectrum.ieee.org/metcalfes-law-is-wrong/](https://spectrum.ieee.org/metcalfes-law-is-wrong/)

# Human-in-the-Loop

Introduction: Human-in-the-loop means deliberately involving people in how an automated system learns, makes decisions, or produces results. A person might label training examples, correct a recommendation, approve an action, or handle an exception the system cannot resolve. The central question is where human judgment adds value and how the product makes that judgment possible. You are designing the division of work between people and technology, including when control changes hands and what happens when they disagree.

**The Need to Know**

Automation can process information and generate options quickly, but it can miss context that changes what a good answer looks like. A suggested customer response might be factually correct yet inappropriate for someone dealing with a repeated service failure. A generated design might look polished while violating a requirement the system never received. Human involvement can bring context, expertise, and accountability to those decisions.

However, adding an approval button does not guarantee meaningful review. The reviewer needs enough information, time, and authority to identify and correct a problem. If the interface hides the original evidence, makes editing difficult, or presents every recommendation as trustworthy, people may simply approve what appears. That creates the appearance of oversight without much benefit. The review experience deserves the same design attention as the automated output.

Human involvement also has a cost. Sending every routine task to a specialist can create delays and make the product expensive to operate. Teams need to decide which actions require review, which exceptions need escalation, and which results can be checked through sampling. Those choices should reflect the consequences of an error, how easily it can be reversed, and the reviewer’s ability to catch it.

**The Real-world Application**

**Adobe Photoshop’s** Generative Fill offers a familiar example of human judgment built into an AI workflow. The user selects an area of an image, provides a prompt or lets the system use the surrounding image as context, and generates content. The user then chooses among the resulting variations. The workflow gives the person control over both the request and the selection of the result.

Consider a designer developing imagery for a product launch. Generative Fill might help explore a different background or add an element to a composition. The designer still needs to assess whether the result matches the brief, represents the product accurately, and works with the rest of the design. This is a real-world use of the feature: the system contributes possibilities, and the person evaluates their suitability.

The product lesson is that human control needs to be specific. Selecting where a change happens, comparing alternatives, and revising a request are useful forms of involvement. Each gives the person a way to influence the outcome. Choosing an output also should not be confused with retraining the underlying model; human input can improve the immediate result without automatically changing how the system behaves in future sessions.

**How this Applies to You**

You should design the handoff as carefully as the automated task. Specify what the system does, when a person becomes involved, what information that person receives, and what actions they can take. “A human will review it” leaves too much unanswered. Name the reviewer, define the decision, and establish what happens if they reject the recommendation or are unavailable.

Evaluate the combined performance of the person and the system. Measure the quality of the final outcome alongside review time, missed errors, and unnecessary escalations. Watch people use the workflow. Can they explain why they accepted a suggestion? Can they correct it without starting over? A high approval rate could reflect strong recommendations, but it could also mean reviewers are rushing or overlooking mistakes.

**Continue Down the Rabbit Hole …**

* **When Combinations of Humans and AI Are Useful** [*Nature Human Behaviour*]
  [https://nature.com/articles/s41562-024-02024-1](https://nature.com/articles/s41562-024-02024-1)

* **Building Effective Agents** [*Anthropic*]
  [https://anthropic.com/engineering/building-effective-agents](https://anthropic.com/engineering/building-effective-agents)

* **Oversee an AI Agent with Human-in-the-Loop** [*IBM*]
  [https://ibm.com/think/tutorials/human-in-the-loop-ai-agent-langraph-watsonx-ai](https://ibm.com/think/tutorials/human-in-the-loop-ai-agent-langraph-watsonx-ai)

* **Lyft Ties Up with Anthropic for AI-Powered Customer Care** [*Reuters*]
  [https://reuters.com/technology/artificial-intelligence/ride-hailing-platform-lyft-ties-up-with-anthropic-ai-powered-customer-care-2025-02-06/](https://reuters.com/technology/artificial-intelligence/ride-hailing-platform-lyft-ties-up-with-anthropic-ai-powered-customer-care-2025-02-06/)

# Generative AI and Agentic Workflows

Introduction: AI can help people interpret information, generate content, and make predictions. Agentic workflows extend those capabilities into action: a system works toward a defined goal by gathering context, choosing appropriate tools, taking permitted steps, and checking the results. For product designers and managers, this expands what a product can do on someone’s behalf. A useful assistant might summarize a support request; an agentic workflow could investigate the issue, make an authorized change, and confirm that it worked. The product challenge is deciding what to delegate, how much discretion to allow, and how users stay informed and in control.

**The Need to Know**

Many products help people move through a process one screen at a time. Users still have to interpret the information, decide what comes next, and transfer details between systems. AI creates opportunities to reduce that coordination work. A customer asking to reschedule a delivery, for example, wants a confirmed new arrival date. An effective workflow could identify the order, check availability, obtain any necessary approval, and update the booking.

Different tasks require different levels of flexibility. A useful distinction is whether the sequence is predefined or whether the model chooses its next step based on what it finds. Terminology varies, but this distinction helps teams avoid giving a system more autonomy than the task requires.

The economics and failure modes also change. An agent may save users several steps while creating additional processing costs, delays, or review work. A mistaken action can affect an account or transaction, making recovery part of the product experience. Teams should evaluate the full task, including corrections and escalations. A convincing conversation is insufficient evidence that the underlying work was completed correctly.

| Approach | How it works | Illustrative product task |
| :---- | :---- | :---- |
| **AI assistance** | Produces information or a draft for someone to use | Summarize customer interviews for a researcher |
| **Defined AI workflow** | Uses AI within a predetermined sequence | Extract details from a request, check required fields, and route it |
| **Agentic workflow** | Selects tools and adjusts steps within assigned boundaries | Investigate an account issue, attempt an allowed fix, and verify the result |

**The Real-world Application**

**Wiley**, the research and education publisher, provides a practical example through its use of Salesforce’s Agentforce. In a published customer case study, the company describes a pilot that used its knowledge base to answer questions and could resolve account access issues and reset passwords. The system also directed registration and payment questions to appropriate resources, leaving more complex issues for service representatives.

The same case study reports more than a 40% improvement in case resolution compared with Wiley’s previous chatbot during the first few weeks. This is a vendor-reported result from that implementation, rather than a general benchmark for AI agents. Its relevance is the connection between conversation and execution: the system could help complete selected processes rather than only supply instructions.

For a team designing a similar service, the key decisions would include how identity is verified, which account changes are allowed, and what happens when an attempted fix fails. Success would mean that the customer can actually regain access. Useful measures would include verified resolution, repeat contacts, time to resolution, and customer satisfaction. Those measures would reveal whether automation reduced the customer’s work or merely moved it elsewhere.

**How this Applies to You**

Begin with a specific task and a clear definition of completion. “Help customers manage their account” leaves too much unresolved. “Restore access after verifying identity, or hand the case to a representative with the relevant context” gives the team something concrete to design and test. Map the information, permissions, and system connections needed to accomplish that task before deciding how much AI it requires.

Design delegation as carefully as interaction. Users need to understand what the system is doing, what it has changed, and when their input is required. Distinguish between suggesting an action, preparing it for review, and carrying it out. Provide a practical way to interrupt, correct, or recover from mistakes. A handoff should preserve useful context so the user does not have to start again.

Finally, evaluate realistic tasks, including incomplete requests, conflicting records, unavailable tools, and unsuccessful actions. Compare the workflow with the existing process on completion quality, total time, cost, and user effort. Expand its authority only when the evidence supports doing so. Your responsibility extends through the outcome, including the work required when the AI gets something wrong.

**Continue Down the Rabbit Hole …**

* **The Cybernetic Teammate: How AI Is Reshaping Collaboration and Expertise in the Workplace** [*Harvard Business School*]
  [https://aiinstitute.hbs.edu/the-cybernetic-teammate-how-ai-is-reshaping-collaboration-and-expertise-in-the-workplace/](https://aiinstitute.hbs.edu/the-cybernetic-teammate-how-ai-is-reshaping-collaboration-and-expertise-in-the-workplace/)

* **4 New Ways to Go From Idea to Product With AI Tools** [*Figma*]
  [https://figma.com/blog/4-new-ways-to-go-from-idea-to-product-with-ai-tools/](https://figma.com/blog/4-new-ways-to-go-from-idea-to-product-with-ai-tools/)

* **How V7 Gives AI Agents Institutional Memory** [*OpenAI*]
  [https://openai.com/index/v7](https://openai.com/index/v7)

#

# Product Positioning [The 4 P’s]

Introduction: Product positioning defines who a product is for, what value it provides, and why someone should choose it over the alternatives. The 4 P’s: Product, Price, Place, and Promotion help turn that positioning into decisions customers can experience. They are traditionally known as the marketing mix. Used together, they connect what you build, what you charge, where you sell, and how you communicate. A clear position becomes credible when all four support the same promise. Calling a product “premium,” “simple,” or “accessible” means little if the buying and usage experience tells a different story.

**The Need to Know**

Customers evaluate the entire offer. A thoughtfully designed product can struggle if it is priced beyond its intended audience’s budget, sold through inconvenient channels, or described in language that misses the reason people need it. The four P’s help designers and managers identify these mismatches before treating every growth problem as a missing feature.

The choices influence one another. A product sold through expert consultation can support a different buying experience from one purchased without assistance. A subscription changes expectations about ongoing value. A promise of simplicity creates obligations for setup, billing, and support. Positioning helps the team decide which combinations make sense for its intended customers.

Good positioning also requires choosing the relevant alternatives. Customers might compare a product with a competitor, a manual process, or doing nothing. A team selling convenience needs to understand what time and effort customers currently spend. A team selling premium quality needs to understand which differences customers actually notice and value.

| Element | What it covers | Positioning question |
| :---- | :---- | :---- |
| **Product** | Functionality, design, quality, packaging, and supporting services | What specific value do we deliver, and what makes it credible? |
| **Price** | What customers pay, how they pay, and the total cost of ownership | Does the price fit the value customers perceive and the economics of serving them? |
| **Place** | Distribution, purchasing channels, availability, and delivery | Can the intended customer discover, buy, and receive the product conveniently? |
| **Promotion** | Messaging, advertising, demonstrations, sales materials, and other communication | Are we explaining a relevant reason to choose us and providing evidence? |

**The Real-world Application**

**Nespresso** illustrates how product and buying experience can reinforce a consistent position. Its founding idea was to let people prepare espresso with the ease of a portioned coffee system. Its offering combines machines and coffee with boutiques, online ordering, and customer assistance. Together, those choices support a proposition that can be understood as quality coffee made convenient.

Viewed through the four P’s, the product makes preparation repeatable and straightforward. Price involves both the machine purchase and subsequent coffee purchases, so the customer evaluates an ongoing experience rather than a single transaction. Place includes boutiques where customers can explore coffees and digital channels where they can reorder. Promotion, including the brand’s George Clooney advertising, adds an aspirational association to the functional promise.

For a product team, the useful lesson is the consistency across those decisions. An attractive machine alone would leave much of the experience unresolved. Customers also need to understand the coffee choices, obtain replacements conveniently, and feel that the experience justifies the ongoing expense. The positioning depends on delivering that complete experience repeatedly.

**How this Applies to You**

You should treat positioning as an input to development. Before debating features, establish the intended customer, the problem that matters, the alternatives they consider, and the reason your offering deserves their attention. A useful working statement is: “For this customer, facing this problem, our product provides this benefit, supported by this evidence.” Specific answers make scope decisions easier.

Work across the four P’s even when you do not own every decision. Designers should understand how pricing and distribution affect customer expectations. Product leaders should understand whether marketing promises are supported by the product. A cheaper version may require simpler onboarding and lower support costs. A premium version may require better materials, greater reliability, or more personal service.

Finally, test whether customers understand and experience the position you intended. Ask what they believe the product is for, which alternatives they considered, and why they chose or rejected it. Compare those answers with purchasing behavior, retention, and support feedback. Your positioning is working when the right customers recognize the value, can access it, and find that the product delivers what they expected.

**Continue Down the Rabbit Hole …**

* **What Is Product Positioning? Definition and Template** [*Shopify*]
  [https://shopify.com/blog/what-is-product-positioning](https://shopify.com/blog/what-is-product-positioning)

* **Crafting Your Brand Identity: A Guide to Positioning Statements and Messaging Pillars** [*University of Virginia Darden School of Business*]
  [https://store.hbr.org/product/crafting-your-brand-identity-a-guide-to-positioning-statements-and-messaging-pillars/UV9447](https://store.hbr.org/product/crafting-your-brand-identity-a-guide-to-positioning-statements-and-messaging-pillars/UV9447)

* **The 4 Ps of Marketing: What They Are and How to Use Them** [*Coursera*]
  [https://coursera.org/articles/4-ps-of-marketing](https://coursera.org/articles/4-ps-of-marketing)

# SWOT Analysis [Framework]

Introduction: SWOT stands for Strengths, Weaknesses, Opportunities, and Threats. It is a framework for evaluating how your team’s capabilities fit the market you want to serve. SWOT helps connect customer needs, competitive conditions, and organizational realities before committing to a product direction. An attractive opportunity does not automatically mean your team is equipped to pursue it. Likewise, a strong technical capability matters only if it helps solve a problem customers care about. A useful SWOT analysis makes those relationships explicit and leads to decisions about where to compete, what to build, and which gaps to address.

**The Need to Know**

Product teams have more possible directions than they have time or resources to pursue. SWOT provides a shared structure for discussing those choices. Its central distinction is between internal factors, your strengths and weaknesses, and external factors, the opportunities and threats in your market.

| Factor | What to examine | Example in product development |
| :---- | :---- | :---- |
| **Strengths** | Internal capabilities or assets that support your objective | Manufacturing expertise that lets you deliver consistent quality at a lower cost |
| **Weaknesses** | Internal limitations that could prevent success | Limited experience supporting customers after a hardware launch |
| **Opportunities** | External conditions your product could benefit from | A growing customer segment whose needs existing products overlook |
| **Threats** | External conditions that could undermine your plans | A competitor introducing a cheaper alternative that meets the same need |

Specificity makes the framework useful. “Great design” is a claim. “Customers completed setup twice as quickly with our prototype as with the leading alternative” is evidence of a potential strength. Similarly, “launch an app” is an action, not an external opportunity. The opportunity might be that customers increasingly need to complete a task while away from their desks. Define the product, target customer, and decision before filling in the matrix.

The next step is to connect the findings. Which strength helps you pursue an opportunity? Which weakness leaves you exposed to a threat? Suppose your team has strong repair expertise, customers are struggling with maintenance, and your distribution budget is limited. That combination could support a focused service offering through existing dealers. SWOT helps generate that option, but customer research and financial analysis still need to establish whether it makes sense.

**The Real-world Application**

**Disney’s** move into streaming illustrates how internal capabilities and external changes can shape a product strategy. In 2017, Disney announced plans to acquire majority ownership of BAMTech while preparing a Disney-branded streaming service. The announcement paired Disney’s content library and brands with BAMTech’s streaming technology, analytics, and commerce capabilities. Disney explicitly connected the investment to a changing media market and the opportunity to build direct relationships with consumers.

A retrospective SWOT interpretation would identify Disney’s content and recognizable characters as strengths, and the need to secure greater control over streaming capabilities as a gap to address. Changing viewing habits created an opportunity to distribute content directly. Competition for viewers’ attention and spending represented a threat.

The resulting decision connected those factors: combine existing content advantages with acquired technology to support a new distribution model. For product teams, this raises a useful question about capability gaps. Should you develop a capability internally, obtain it through a partner or acquisition, or choose a narrower product scope? Identifying a weakness becomes valuable when it changes how you plan to deliver the product.

**How this Applies to You**

Evaluate your idea in relation to the organization that must deliver it. A compelling concept may depend on capabilities your team does not yet have. An established strength may offer little advantage in a new market. Ask what customers would actually notice and value, and compare your claims with the alternatives available to them.

Make disagreement useful. Invite engineering, design, sales, operations, and support to challenge the analysis with evidence. Keep assumptions visible, rank findings by their relevance to the decision, and avoid treating a longer list of strengths as proof that the idea deserves funding. One major constraint can outweigh several modest advantages. Always remember to ask, “Given what we know, what are we going to do differently?”

**Continue Down the Rabbit Hole …**

* **SWOT Analysis** [*Nielsen Norman Group*]
  [https://nngroup.com/articles/swot-analysis/](https://nngroup.com/articles/swot-analysis/)

* **SWOT Analysis: How To Do One (With Template & Examples)** [*HubSpot*]
  [https://blog.hubspot.com/marketing/swot-analysis/](https://blog.hubspot.com/marketing/swot-analysis/)

* **What Is a SWOT Analysis?** [*Coursera*]
  [https://coursera.org/articles/swot-analysis/](https://coursera.org/articles/swot-analysis/)

* **SWOT Analysis: Examples and Templates** [*Asana*]
  [https://asana.com/resources/swot-analysis](https://asana.com/resources/swot-analysis)

Work SWOT Analysis Template by *ClickUp* A customizable ClickUp template for mapping strengths, weaknesses, opportunities, and threats, helping your team assess its position and turn insights into a strategic action plan.[https://clickup.com/templates/swot-analysis/work](https://clickup.com/templates/swot-analysis/work)

# Backcasting [Framework]

Introduction: Backcasting starts with a desired future and works backward to identify what needs to happen to get there; connecting an ambitious product vision to the work your team should do next. Define what success looks like for the customer, choose a time horizon, and ask: “What would need to be true before this could happen?” Repeat that question until you reach decisions you can make today. The result is a sequence of requirements, milestones, and assumptions you can test.

**The Need to Know**

Product teams spend much of their time responding to what is immediately in front of them: customer requests, competitive pressure, technical issues, and quarterly targets. Those demands matter, but they can consume the entire roadmap. A team can deliver improvements every month without getting closer to the experience it ultimately wants to create. Backcasting gives those immediate decisions a longer-term purpose.

Forecasting asks what is likely to happen based on current conditions and trends. Backcasting asks what would need to happen to reach a chosen outcome. Both are useful. Backcasting is especially helpful when the desired future requires meaningful changes to the product, business model, or supporting infrastructure. The UK Government Office for Science recommends it for identifying the steps toward a preferred future, including dependencies outside the team’s control.

Consider a team that wants customers to install a home appliance in ten minutes without professional help. Working backward reveals more than a need for better instructions. The product may require different connectors, fewer assembly steps, clearer packaging, and a way to verify that installation was successful. That changes what the team needs to prototype now. A useful future outcome exposes decisions that a feature-by-feature roadmap might not.

**The Real-world Application**

Tesla’s original 2006 Master Plan provides a useful illustration of backcasting logic. The stated direction was to make electric vehicles accessible to a broader market. The proposed sequence began with a sports car, then used proceeds to support a more affordable car, followed by an even more affordable model. The initial product was intended to help enable the eventual market ambition.

Read backward, the plan raises a series of dependencies: broader adoption requires affordability; affordability requires lower costs and greater production capability; developing those capabilities requires resources and an initial market willing to pay more. It also describes the proposed strategy, not proof that early vehicle sales alone financed its execution.

For product teams, the useful lesson is to explain how an early release makes a later outcome possible. Your first product might serve a narrower customer group, establish a critical capability, or test a difficult assumption. Be specific about what it contributes. “We’ll start small and expand” is incomplete unless you can explain what the first step teaches/enables.

**How this Applies to You**

You should become more precise about the future you are asking the team to build. “A better customer experience” gives people very little direction. “Within three years, a first-time customer can complete the setup in ten minutes without contacting support” gives the team something it can investigate. Designers can make that future experience tangible through prototypes and scenarios. Product owners can connect it to customer demand, economics, and organizational capabilities.

Then work backward together. Identify the conditions required for that experience, the milestones that would demonstrate progress, and the assumptions that could invalidate the plan. Distinguish what your team controls from what depends on suppliers, partners, or changes in customer behavior. Give the nearest milestones owners and evidence requirements, so the exercise leads to action.

Treat the resulting path as something to test and revise. A compelling future does not establish that customers want it or that the business can deliver it. Think through the assumption that most threatens a plan and the smallest useful test you can run now. Backcasting earns its place when it changes today’s priorities.

**Continue Down the Rabbit Hole …**

* **Backcasting from the Future** [*UC Berkeley*]
  [https://cmr.berkeley.edu/2024/03/backcasting-from-the-future-strategies-for-accelerating-and-de-risking-discontinuous-innovations/](https://cmr.berkeley.edu/2024/03/backcasting-from-the-future-strategies-for-accelerating-and-de-risking-discontinuous-innovations/)

* **The Futures Toolkit: Backcasting** [*UK Government Office for Science*]
  [https://gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts/the-futures-toolkit-html#backcasting](https://gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts/the-futures-toolkit-html#backcasting)

* **Digital Engineering Needs Through Human-Centered Approaches** [*AIAA*]
  [https://doi.org/10.2514/6.2025-2033](https://doi.org/10.2514/6.2025-2033)

Working backwards: Backcasting Template by *Mural* A customizable Mural template for defining a future goal and working backward to map the steps needed to achieve it, helping your team turn a shared vision into an actionable plan.[https://mural.co/templates/working-backwards/](https://mural.co/templates/working-backwards/)

# Kano Model [Framework]

Introduction: Product features affect customer satisfaction in different ways. The Kano Model explores how basic expectations, performance improvements, and unexpected delights shape product decisions; helping teams prioritize features, balance investment, and recognize how customer expectations evolve over time.

**The Need to Know**

Every product team has more ideas than time, money, or people to execute them. Customers request features. Sales wants something that will close the next deal. Engineering sees opportunities to improve performance. Leadership wants something competitors don't have. You need a way to evaluate what those investments will actually mean to the customer.

The Kano Model separates customer responses into three main categories:

* **Basic needs**: Customers expect these to work. Meeting the expectation earns little credit, but falling short creates dissatisfaction. Think of a checkout process that charges the correct amount.

* **Performance needs**: Better performance produces greater satisfaction. Think of battery life, delivery speed, or the time it takes to complete a task.

* **Delighters**: Customers appreciate these even though they didn't expect them. Their absence is acceptable, but their presence can make the product stand out.

Two additional categories matter: **indifferent** features have little effect on satisfaction, while **reverse** features make some customers less satisfied when included. A feature your team loves can fall into either category. These classifications also change as expectations evolve.

The practical implication is that effort and value aren't proportional. Six months spent building something impressive can matter less than fixing a frustrating part of an everyday workflow. Likewise, meeting every basic expectation may still leave customers with no compelling reason to choose you over the competition. Start with a specific customer segment and a short list of potential features. For each feature, ask two questions: How would you feel if the product included it? How would you feel if it didn't?

**The Real-world Application**

**Spotify’s** Discover Weekly, introduced in 2015, gives listeners a personalized playlist refreshed every Monday, helping them find music based on their listening habits. It added a new benefit to the experience: customers could discover songs they might enjoy without having to search for them.

Viewed through Kano, reliable playback would be a basic need: listeners expect music to play correctly and become frustrated when it doesn't. Faster loading would be a performance need, with shorter waits improving the experience. For a listener who expected to find their own music, Discover Weekly could be a delighter, offering something valuable they hadn't expected. They also depend on the customer: someone who joins specifically for personalized discovery may consider recommendations a requirement from day one.

**How this Applies to You**

As a product leader, you are responsible for understanding what customers will value and why. Shipping a feature is only part of that responsibility. You should be able to explain the expectation it addresses, whose experience it improves, and what evidence supports the investment.

This also means getting comfortable with work that won't make the launch announcement. Reliability, clear instructions, and fewer errors may deserve more attention than the feature everyone wants to demo. At the same time, a product that only meets expectations may struggle to attract customers.

Before committing to a feature, ask: What happens if we leave this out, and what changes if we get it right? The difference between those answers can tell you a great deal about where your team should spend its time.

**Continue Down the Rabbit Hole …**

* **Lots of Ideas, but Not Sure How to Prioritize?** The Kano Model [*Microsoft Research*]
  [https://microsoft.com/en-us/research/articles/lots-of-ideas-but-not-sure-how-to-prioritize-the-kano-model-can-help/](https://microsoft.com/en-us/research/articles/lots-of-ideas-but-not-sure-how-to-prioritize-the-kano-model-can-help/)

* **How to Use the Kano Model to Prioritize Features** [*ProductPlan*]
  [https://productplan.com/learn/kano-model-prioritization](https://productplan.com/learn/kano-model-prioritization)

* **Improving Consumer Satisfaction:** A Kano-Based Approach [*Elsevier Journal*]
  [https://www.sciencedirect.com/science/article/pii/S0969698925000220](https://www.sciencedirect.com/science/article/pii/S0969698925000220)

* **The Kano Model: A Tool to Prioritize the Users' Wants and Desires** [*Interaction Design Foundation*]
  [https://interaction-design.org/literature/article/the-kano-model-a-tool-to-prioritize-the-users-wants-and-desires](https://interaction-design.org/literature/article/the-kano-model-a-tool-to-prioritize-the-users-wants-and-desires)

Kano Model Template by *Creately* Use Creately’s Kano Model Generator to transform messy feedback into a clear product roadmap by visualizing satisfaction levels and identifying high-impact features.[https://creately.com/usage/kano-model/](https://creately.com/usage/kano-model/)

# R.I.C.E. [Framework]

Introduction: RICE stands for Reach, Impact, Confidence, and Effort. It is a prioritization framework that helps product teams compare initiatives based on how many people they will affect, how much those people will benefit, how strong the supporting evidence is, and how much work delivery requires. The formula is RICE Score = (Reach × Impact × Confidence) ÷ Effort. Its value comes from making assumptions visible so a team can discuss tradeoffs and decide where to invest its limited resources.

**The Need to Know**

Product teams usually have more worthwhile ideas than they can deliver. Customer requests, usability problems, sales commitments, and new opportunities all compete for attention. Without a shared way to evaluate them, priorities can depend on who argues most persuasively or which problem was raised most recently.

These are the four inputs in Intercom’s original framework.

| Factor | What it measures | How to estimate it |
| :---- | :---- | :---- |
| **Reach** | People or events affected within a defined period | Use the same unit and timeframe across initiatives. |
| **Impact** | Expected effect on each person against a chosen goal | Use a shared scale, such as 0.25, 0.5, 1, 2, or 3. |
| **Confidence** | Strength of the evidence behind the estimates | Express as a percentage, entered as a decimal in the formula. |
| **Effort** | Total work required across the team | Estimate in person-months, including design, product, and engineering. |

The quality of the discussion depends on the quality of the inputs. Define the desired outcome before scoring, whether that is successful onboarding, repeat usage, or another customer behavior. A feature that many people encounter may have little effect on that outcome. Likewise, a confident presentation does not justify a high confidence score. Ask what evidence supports each estimate and whether reasonable changes to those estimates would change the ranking.

**The Real-world Application**

**Intercom**, the customer service software company, developed RICE to compare competing product ideas more consistently. In its published explanation, the company describes combining the four factors into a score, reviewing the resulting order, and reconsidering estimates that appear questionable. It also acknowledges that dependencies or essential customer requirements can justify prioritizing a lower-scoring initiative.

Consider a hypothetical customer service product team trying to increase the number of customers who successfully complete setup. The following figures illustrate how that team might apply RICE; they are not reported Intercom project results.

| Initiative | Reach per quarter | Impact | Confidence | Effort in person months | RICE score |
| :---- | :---- | :---: | :---: | :---: | :---: |
| **Simplify the setup checklist** | 2,000 customers | 1 | 80% | 2 | 800 |
| **Add guided data import** | 800 customers | 2 | 80% | 4 | 320 |
| **Build a new setup assistant** | 2,000 customers | 3 | 50% | 8 | 375 |

The checklist ranks first because its combination of reach, confidence, and effort outweighs its lower estimated impact per customer. The assistant has greater potential impact, but also requires more work and rests on weaker evidence. That could justify testing a prototype before committing to development. The import feature could still take precedence if customers cannot complete setup without it. The score makes the tradeoff easier to examine.

**How this Applies to You**

Treat prioritization as an argument you should be able to explain. For each proposal, identify who benefits, what should improve, what evidence supports the claim, and what delivery requires. Bring design, engineering, research, and business perspectives into the estimates. A roadmap built from one person’s assumptions can look precise while missing substantial work or customer needs.

Use uncertainty to guide discovery. If an idea ranks poorly because its impact is unknown, identify the smallest useful study or experiment that could resolve that uncertainty. Also examine whether the scoring process consistently favors familiar improvements over unfamiliar opportunities. You may need to reserve capacity for exploration, infrastructure, and essential maintenance rather than making every investment compete in one ranked list.

**Continue Down the Rabbit Hole …**

* **RICE: Simple prioritization for product managers** [*Intercom*]
  [https://intercom.com/blog/rice-simple-prioritization-for-product-managers/](https://intercom.com/blog/rice-simple-prioritization-for-product-managers/)

* **How to Use the RICE Framework for Better Prioritization** [*Product School*]
  [https://productschool.com/blog/product-fundamentals/rice-framework](https://productschool.com/blog/product-fundamentals/rice-framework)

* **Guide: Mastering prioritization with RICE** [*Atlassian*]
  [https://community.atlassian.com/forums/Jira-Product-Discovery-articles/Guide-Mastering-prioritization-with-RICE/ba-p/3169901](https://community.atlassian.com/forums/Jira-Product-Discovery-articles/Guide-Mastering-prioritization-with-RICE/ba-p/3169901)

* **How to streamline projects using the RICE prioritization method** [*Notion*]
  [https://notion.com/blog/rice-prioritization](https://notion.com/blog/rice-prioritization)

RICE Template for Effective Prioritization by *ProductLift* A collection of free RICE prioritization templates available in various formats. All of the RICE prioritization templates are designed to be: easy-to-use, customizable, and collaborative.[https://productlift.dev/blog/rice-template/](https://productlift.dev/blog/rice-template/)

# Strategic Sourcing [Framework]

Introduction: Strategic sourcing is the process of deciding which suppliers, materials, technologies, and services will best support your product and business over time. It considers price alongside quality, reliability, capability, and risk. The key idea is that sourcing decisions are product decisions; the component you select, the manufacturer you work with, or the software service you depend on can determine what your product does, what it costs, and whether you can deliver it consistently. Those choices belong in the development process from the beginning.

**The Need to Know**

A product concept eventually has to work within the capabilities of the organizations helping you deliver it. A custom enclosure might look great but require expensive tooling and a large minimum order. A lower-cost component might have inconsistent quality or a replacement lead time that puts production at risk. A software provider might make your first release easier while creating substantial costs as usage grows. Each sourcing choice affects the customer experience and the economics behind it.

This is why teams need to understand the total cost of ownership. The quoted price is one part of the cost. Shipping, integration, testing, defects, inventory, maintenance, and eventually switching suppliers can change the calculation. Imagine two manufacturers quoting the same part. One offers a lower unit price but requires six months of inventory upfront. The other charges more per unit but supports smaller orders and faster design changes. For a team still learning what customers want, that flexibility may justify the premium.

Suppliers can also contribute expertise your team does not have. A materials partner may suggest a more durable formulation. A manufacturer may identify a simpler assembly method. Bringing those partners into development early can expand your options and expose problems before the design becomes expensive to change. The relationship itself can become a source of product advantage.

**The Real-world Application**

**Patagonia’s** partnership with Yulex shows how sourcing can enable a different product. Patagonia wanted an alternative to conventional neoprene for its wetsuits. In 2016, it introduced a full line of neoprene-free wetsuits using natural rubber developed through that partnership, expanding its Yulex offering from two to 21 full-length suits. The rubber came from sources certified by the Forest Stewardship Council.

The sourcing challenge included the origin of the material. Natural rubber can come from plantations associated with deforestation, so selecting a plant-based input alone would not meet Patagonia’s objective. Its announcement explains that the company waited to use hevea rubber until it found a more sustainable source. The formulation also included synthetic rubber to meet durability and resistance requirements. Material selection, supplier collaboration, traceability, and product performance were connected decisions.

The core objective is to be able to translate the product promise into supplier requirements. If your promise includes durability, repairability, affordability, or responsible materials, identify what evidence a supplier must provide. Then test the resulting product. A supplier’s credentials can support your decision, but the finished experience still has to deliver for the customer.

**How this Applies to You**

You must take an active role in defining what the team needs from its partners. Designers need to understand how materials, manufacturing processes, and component availability affect their choices. Product managers need to connect supplier capabilities to margins, launch timing, and future growth. Work with procurement and engineering early enough that their input can influence the design.

Make the tradeoffs explicit. Agree on essential requirements, compare qualified suppliers using consistent criteria, and validate important claims through samples, prototypes, or pilots. Consider how much dependence you are creating. A specialized supplier may offer a valuable capability, but the team should understand what happens if that supplier raises prices, misses deliveries, or discontinues a component. A backup source can help, although qualifying and maintaining it also takes work. Simply ask, “What are we committing ourselves to by choosing this partner?” That question connects a purchasing decision to the product you will be able to build and support.

**Continue Down the Rabbit Hole …**

* **Transforming Procurement Functions for an AI-Driven World** [*McKinsey*]
  [https://mckinsey.com/capabilities/operations/our-insights/transforming-procurement-functions-for-an-ai-driven-world](https://mckinsey.com/capabilities/operations/our-insights/transforming-procurement-functions-for-an-ai-driven-world)

* **Revitalizing Sourcing to Balance Cost and Resilience** [*BCG*]
  [https://bcg.com/publications/2024/revitalizing-global-sourcing-to-balance-cost-and-resilience](https://bcg.com/publications/2024/revitalizing-global-sourcing-to-balance-cost-and-resilience)

* **Creating an Effective Sourcing and Supply Chain Decision Tool** [*MIT Sloan*]
  [https://mitsloan.mit.edu/action-learning/creating-effective-sourcing-and-supply-chain-decision-tool](https://mitsloan.mit.edu/action-learning/creating-effective-sourcing-and-supply-chain-decision-tool)

* **4 Strategic Sourcing Use Cases** [*IBM*]
  [https://ibm.com/think/topics/sourcing-examples/](https://ibm.com/think/topics/sourcing-examples/)

Sourcing Strategy Template by *Cascade* Develop a comprehensive sourcing strategy with this Google Sheets template to reduce costs and increase competitiveness.[https://cascade.app/google-sheet-templates/sourcing-strategy-template](https://cascade.app/google-sheet-templates/sourcing-strategy-template)

#

# MSCW Method [Framework]

Introduction: MSCW *(also known as MoSCoW)* helps teams decide what a product must include, what it should include, what it could include, and what it won’t include this time. It makes priorities explicit when time, money, and people are limited. A good product concept needs a realistic delivery plan; you need to explain which requirements are essential to the customer’s experience, which can wait, and why.

**The Need to Know**

Product teams rarely struggle to find more things to build. They struggle to agree on what deserves their attention first. Customers request improvements, sales teams make commitments, engineers identify dependencies, and designers see opportunities to improve the experience. Each request can sound reasonable on its own. Together, they can create a product that takes too long to deliver.

MoSCoW gives the team a shared language for those decisions:

| Priority | What it means for this release |
| :---- | :---- |
| **Must Have** | Essential for the release to meet its purpose or be acceptably delivered |
| **Should Have** | Important, but a temporary workaround or delay is tolerable |
| **Could Have** | Useful, with less impact if omitted. Include it if capacity allows |
| **Won’t Have** | Explicitly excluded from the agreed scope *(for now)* |

These categories apply to a defined delivery period. A requirement can be a Must for the eventual product and a Won’t for an early prototype.

The discipline comes from defending the labels. If every requirement is a Must, the team has no room to respond when something takes longer than expected. Ask what would actually happen if a requirement were missing. Would customers be unable to complete the core task? Would they have an acceptable alternative? Those answers are more useful than how strongly someone argues for a feature.

**The Real-world Application**

**IKEA’s** Democratic Design considers five dimensions: function, form, quality, sustainability, and low price. IKEA explains that the balance between these dimensions varies across products. That creates a practical design challenge: improve the product while keeping it appropriate for the people buying it.

We can apply MoSCoW to an illustrative IKEA-style brief for an affordable desk. Suppose the target customer needs a dependable workspace for a small apartment, within a firm price limit. The priorities might look like this:

| Priority | Product features |
| :---- | :---- |
| **Must Have** | A stable, safe work surface that fits the intended space and target price |
| **Should Have** | Cable management that helps keep the workspace organized |
| **Could Have** | Additional color choices, if production capacity permits |
| **Won’t have** | Motorized height adjustment, given the release’s price & production constraints |

**How this Applies to You**

Our job as product leaders includes making the tradeoffs understandable. A designer should be able to explain which parts of an experience protect the product’s core value. A product owner should be able to connect those requirements to customer evidence, business goals, and delivery constraints. Both need to challenge a priority that rests only on someone’s preference.

Use MSCW as a conversation supported by evidence. Define the customer, intended outcome, release date, and available capacity before assigning categories. Bring engineering and other delivery partners into the discussion so that dependencies and effort inform the choices. Revisit priorities when research changes what you know, and make any resulting scope changes explicit.

Simply ask, “If we could deliver only the Musts, would the customer still receive something worth using?” If the answer is no, reconsider the scope. If the answer is yes, you have a clearer basis for deciding which improvements deserve the next investment.

**Continue Down the Rabbit Hole …**

* **What Is MoSCoW Prioritization** [*Agile Business Consortium*]
  [https://agilebusiness.org/resource/what-is-moscow-prioritization/](https://agilebusiness.org/resource/what-is-moscow-prioritization/)

* **MoSCoW Prioritization** [*ProductPlan*]
  [https://productplan.com/glossary/moscow-prioritization](https://productplan.com/glossary/moscow-prioritization)

* **Making Your UX Life Easier with the MoSCoW** [*Interaction Design Foundation*]
  [https://ixdf.org/literature/article/making-your-ux-life-easier-with-the-moscow](https://ixdf.org/literature/article/making-your-ux-life-easier-with-the-moscow)

* **Six Product Prioritization Frameworks and How to Pick the Right One** [*Atlassian*]
  [https://atlassian.com/agile/product-management/prioritization-framework](https://atlassian.com/agile/product-management/prioritization-framework)

MoSCoW Matrix Template by *Miro* A customizable Miro board for sorting product features into Must Have, Should Have, Could Have, and Won’t Have categories, helping your team agree on priorities and project scope.[https://miro.com/templates/moscow-matrix-basic/](https://miro.com/templates/moscow-matrix-basic/)

# PESTLE [Framework]

Introduction: PESTLE stands for Political, Economic, Social, Technological, Legal, and Environmental. It is a framework for examining external forces that could affect a product’s success. PESTLE helps connect the product you want to build with the conditions in which it will actually operate. Customers might love your concept, but changing costs, new requirements, or missing infrastructure could make it difficult to deliver. Those same changes can also create opportunities.

**The Need to Know**

Product teams spend considerable time studying users and competitors. They also need to understand the broader conditions shaping demand, feasibility, and cost. A product can solve a real problem and still arrive too early, cost too much, or depend on infrastructure that customers cannot access. PESTLE gives the team six areas to investigate.

| Factor | What to examine | A product question it raises |
| :---- | :---- | :---- |
| **Political** | Government priorities, trade policies, and political stability | Could a change in trade policy disrupt our sourcing strategy? |
| **Economic** | Purchasing power, inflation, financing costs, and exchange rates | Will our target customers still be able to afford this? |
| **Social** | Demographics, cultural expectations, habits, and attitudes | Are changing routines creating a need our product could address? |
| **Technological** | Emerging capabilities, infrastructure, and technical maturity | Is the technology reliable and accessible enough for this use? |
| **Legal** | Applicable laws, product requirements, and obligations | What requirements must the design satisfy before launch? |
| **Environmental** | Resource availability, waste, emissions, and physical climate risks | Could material scarcity or extreme weather affect delivery? |

The categories can overlap. Government action on waste, for example, may influence both legal requirements and material costs. Focus on the implications rather than debating which box something belongs in. “Inflation” is too broad to guide a team. “Higher component costs could push our retail price beyond what customers will pay” identifies a specific assumption you can investigate.

A useful analysis ends with priorities. Identify the external changes that could most affect your product, estimate when they might matter, and distinguish established facts from uncertain developments. Then decide whether to change the design, test an assumption, find a partner, or monitor a specific signal. Filling six boxes without changing a decision adds little value.

**The Real-world Application**

Maersk’s introduction of the ‘Laura Mærsk’ in 2023 provides a useful example. The logistics company introduced the world’s first methanol-enabled container vessel as part of its effort to reduce shipping emissions. Delivering that service required more than developing a ship: it also required access to an appropriate fuel supply.

Before the vessel entered service, Maersk announced an agreement securing green methanol for its initial months of operation, with refueling in Rotterdam. The announcement also described certification and traceability arrangements for the fuel. These details show how an environmental objective becomes a set of technical, supply, and verification requirements.

Viewed through PESTLE, a team evaluating this service would examine policy direction, fuel economics, customer expectations, engine readiness, applicable requirements, and lifecycle emissions. For product designers and managers, the lesson is practical: a promising technology becomes a viable offering only when the surrounding conditions support its use.

**How this Applies to You**

Take responsibility for understanding the assumptions outside your team’s control. When someone says a product will become cheaper, customers will change their behavior, or infrastructure will become available, ask what evidence supports that expectation. Bring in colleagues from operations, sourcing, finance, and legal when their knowledge could change the design or launch plan.

Make the analysis specific to a market and a time horizon. The conditions for launching next year in one country may differ substantially from those for expanding internationally five years later. Record what would trigger a reassessment, such as a supplier price threshold, a confirmed policy change, or a technology reaching an agreed level of reliability. Keep returning to one question: “What has to be true in the outside world for this product to work?”

**Continue Down the Rabbit Hole …**

* **PESTLE Analysis** [*CIPD*]
  [https://cipd.org/en/knowledge/factsheets/pestle-analysis-factsheet/](https://cipd.org/en/knowledge/factsheets/pestle-analysis-factsheet/)

* **PEST Analysis: Definition, Steps, Examples + PESTLE** [*Asana*]
  [https://asana.com/resources/pest-analysis](https://asana.com/resources/pest-analysis)

* **PEST Analysis: What It Is and How To Conduct One** [*Shopify*]
  [https://shopify.com/blog/pest-analysis/](https://shopify.com/blog/pest-analysis/)

* **What Is PESTLE Analysis & How Does It Work?** [*Atlassian*]
  [https://atlassian.com/work-management/strategic-planning/pestle-analysis/](https://atlassian.com/work-management/strategic-planning/pestle-analysis/)

PESTLE analysis templates by *Canva* A collection of customizable Canva whiteboard templates for mapping political, economic, social, technological, legal, and environmental factors.[https://canva.com/online-whiteboard/templates/pestle-analysis/](https://canva.com/online-whiteboard/templates/pestle-analysis/)

#

# RFM [Framework]

Introduction: RFM stands for Recency, Frequency, and Monetary Value. It is a framework for grouping customers based on how recently they purchased, how often they purchase, and how much they spend. RFM provides a practical way to connect customer behavior with product decisions; someone who made their first purchase yesterday has a different relationship with your product than someone who bought regularly for years but recently stopped. Understanding those differences helps you decide which experiences to improve, which customers to learn from, and where to investigate problems. RFM describes patterns in past behavior; you still need research to understand what caused them.

**The Need to Know**

An average can hide the very problem your team needs to solve. Revenue might look healthy because new customers are spending more, even as regular customers stop returning. RFM helps separate these patterns so you can ask more useful questions about retention, repeat purchases, and customer value.

| Dimension | What it measures | What it helps you investigate |
| :---- | :---- | :---- |
| **Recency** | Time since the customer’s last purchase | Is this customer still buying, or has their usual pattern changed? |
| **Frequency** | Number of purchases within a defined period | Are customers returning after their first purchase? |
| **Monetary Value** | Total customer spending within that period | Which customers account for more revenue? |

A common approach scores each dimension from 1 to 5, with higher scores representing more recent purchases, more frequent purchases, and greater spending. These scores help form groups such as new customers, regular buyers, and previously valuable customers who have become inactive. Set the measurement period and thresholds around the business’s buying cycle. Six months without a purchase means something very different for groceries than for furniture.

For designers and managers, those groups can guide research and product improvements. First-time buyers might need better setup instructions. Repeat buyers might benefit from easier reordering. Previously frequent buyers might reveal a service problem that aggregate metrics missed. Spending alone does not tell you profitability, satisfaction, or future potential, and a new customer’s low frequency may simply reflect how recently they joined.

**The Real-world Application**

**Shopify** has built RFM analysis into its merchant reporting tools. Its customer analysis assigns scores across the three dimensions and organizes customers into 11 groups. The scores are relative to customers within that merchant’s store, rather than an industry benchmark. This gives merchants a practical way to distinguish different purchasing relationships and decide where to focus their attention.

Consider a hypothetical skincare merchant using these tools. One customer purchased yesterday for the first time. Another purchases every month. A third previously purchased frequently and spent heavily, but has not returned in several months. Sending all three the same promotion would overlook meaningful differences. The first might benefit from guidance on using the product, the second from a simpler replenishment experience, and the third from a conversation about why they stopped buying.

The product opportunity comes from investigating those differences. If interviews with previously frequent buyers uncover difficulty finding compatible refills, the team could improve product compatibility information and the reorder flow. It could then test whether those changes increase repeat purchases. RFM identifies a group worth investigating; the team still has to establish the problem and test the solution.

**How this Applies to You**

Be more specific about whom you are designing for. “Our customers want this” is often too broad to support a decision. Ask which customers, what their purchasing relationship looks like, and whether the proposed change addresses a problem they actually have. Use RFM segments to recruit research participants across different behaviors, including people who stopped returning.

Treat each segment as a starting point for inquiry. A declining recency score could reflect dissatisfaction, seasonality, a durable product that needs no replacement, or a change in the customer’s circumstances.

**Continue Down the Rabbit Hole …**

* **What Is RFM Analysis? Definition, Benefits & Examples** [*CleverTap*]
  [https://clevertap.com/blog/rfm-analysis/](https://clevertap.com/blog/rfm-analysis/)

* **RFM Analysis (Recency, Frequency, Monetary)** [*Informa TechTarget*]
  [https://techtarget.com/data-technologies/definition/RFM-analysis-recency-frequency-monetary](https://techtarget.com/data-technologies/definition/RFM-analysis-recency-frequency-monetary)

* **What Is RFM?** [*Intuit Mailchimp*]
  [https://mailchimp.com/resources/rfm-analysis/](https://mailchimp.com/resources/rfm-analysis/)

* **RFM Segmentation** [*Optimove*]
  [https://optimove.com/resources/learning-center/rfm-segmentation](https://optimove.com/resources/learning-center/rfm-segmentation)

RFM Analysis Template by *DataHub Pro* An interactive template with Excel export for segmenting customers by purchase recency, frequency, and monetary value.[https://datahubpro.co.uk/templates/rfm-analysis/](https://datahubpro.co.uk/templates/rfm-analysis/)

#

# RACI Matrix

Introduction: A RACI Matrix clarifies who does the work, who owns the outcome, whose input is needed, and who needs an update. The letters stand for Responsible, Accountable, Consulted, and Informed. The matrix helps translate a shared product goal into clear individual commitments as product development requires collaboration across disciplines, but collaboration alone does not establish ownership. RACI makes those expectations visible before missed handoffs, competing assumptions, or unclear approvals slow the team down.

**The Need to Know**

Product teams can agree on what they want to accomplish while holding different assumptions about who will make it happen. A designer may expect the product manager to arrange customer interviews. The product manager may assume the designer already scheduled them. Engineering may begin implementation before someone has confirmed the requirements. These problems often come from gaps in coordination, even when everyone is capable and committed.

RACI separates four types of involvement:

| Role | What it means  |
| :---- | :---- |
| **Responsible** | Performs the work. More than one person can contribute to execution |
| **Accountable** | Owns completion and the outcome. Assign one person per task or deliverable |
| **Consulted** | Provides expertise or input before the work or decision is finalized. Communication goes both ways |
| **Informed** | Receives relevant updates about progress and outcomes. Participation or approval is generally unnecessary |

The same person can be both Responsible and Accountable. The distinction is between doing the work and ensuring that it reaches an acceptable result.

A matrix places tasks or deliverables in rows and people or roles in columns, with letters marking each assignment. Reviewing it exposes missing owners, overloaded teammates, and unnecessary review steps. Keep it focused on work where coordination matters. A matrix with hundreds of tiny tasks can become harder to maintain than the project itself, and assigning someone accountability does little unless they have the authority and capacity to fulfill it.

**The Real-world Application**

**GitLab** provides a documented example of making ownership explicit. Its public handbook describes Directly Responsible Individuals, or DRIs, who own particular projects, initiatives, or activities. It connects that approach to RACI by placing the DRI in the Accountable role for a given project or workstream. The handbook also provides a RACI template for teams starting new work.

GitLab gives a product development example: the product manager owns prioritization, while the engineering manager owns delivery. Ownership therefore depends on the activity. The company also expects owners to consult relevant teams and gather their expertise. A clear owner can make progress while still drawing on the people needed to make a sound decision.

For a classroom application, imagine a team redesigning onboarding. A designer could be Responsible for the usability study, a research lead Accountable for its quality and completion, the product manager and engineer Consulted on the study questions, and customer support Informed of the findings. Its value comes from agreeing on these expectations before work begins.

**How this Applies to You**

You should treat clarity about ownership as part of your work. At the start of an initiative, identify the deliverables most likely to create confusion and agree on the roles together. Avoid automatically making the product owner Accountable for every row. Ownership should reflect the work, the expertise required, and the authority to move it forward.

Be deliberate about consultation. Someone marked Consulted needs a real opportunity to contribute before the relevant decision is settled. Someone marked Informed needs a useful update without being pulled into every meeting. Establish when input is due and how disagreements will be resolved. The letters alone cannot answer those questions, and consultation does not automatically give every participant veto power.

**Continue Down the Rabbit Hole …**

* **Your Guide to RACI Charts, with Examples** [*Asana*]
  [https://asana.com/resources/raci-chart](https://asana.com/resources/raci-chart)

* **What Is a RACI Chart? Definition, Benefits, and Examples** [*Wrike*]
  [https://wrike.com/blog/what-is-a-raci-chart/](https://wrike.com/blog/what-is-a-raci-chart/)

* **Align Responsibilities Across Teams** [*Microsoft*]
  [https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/organize/raci-alignment/](https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/organize/raci-alignment/)

* **RACI Chart: What Is It & How to Use It** [*Atlassian*]
  [https://atlassian.com/work-management/project-management/raci-chart/](https://atlassian.com/work-management/project-management/raci-chart/)

Excel RACI Template by *TeamGantt*  A customizable Excel template for assigning Responsible, Accountable, Consulted, and Informed roles to project tasks.[https://teamgantt.com/blog/responsibility-assignment-matrix-template/](https://teamgantt.com/blog/responsibility-assignment-matrix-template/)

# New’ish & Emerging Technologies

Overview: Innovation and disruption from our industry peers is an accelerant for product leaders to have the earliest opportunity to explore where the world might be heading and help shape what comes next. Being an early adopter means looking beyond the novelty of a technology to understand what it makes possible, which customer problems it could solve, and how it might change people’s expectations or behavior. Experimenting before the path is obvious helps teams challenge assumptions, build practical knowledge, and recognize opportunities before they become conventional wisdom. The goal is not to predict every winner, but to learn faster through small, deliberate experiments that distinguish what is possible from what is useful. Product leaders who develop this habit are better prepared to turn technological change into products people actually want.

**Generative AI / Agentic AI**

Generative AI creates content such as text, images, and code, while agentic AI uses models and tools to take actions toward a goal across multiple steps. For product leaders, these technologies create opportunities to accelerate development and design products that perform work on a user’s behalf, with clear boundaries around autonomy, accuracy, and human oversight.

* **Disney invests in OpenAI and licenses characters for Sora** [*Reuters*]
  [https://reuters.com/business/media-telecom/disney-makes-1-billion-investment-openai-brings-characters-sora-2025-12-11/](https://reuters.com/business/media-telecom/disney-makes-1-billion-investment-openai-brings-characters-sora-2025-12-11/)

* **Starbucks pilots a generative AI assistant for baristas** [*Investopedia*]
  [https://investopedia.com/starbucks-is-piloting-a-genai-virtual-assistant-for-baristas-11752338](https://investopedia.com/starbucks-is-piloting-a-genai-virtual-assistant-for-baristas-11752338)

* **McDonald’s gives its restaurants an AI makeover** [*The Wall Street Journal*]
  [https://wsj.com/articles/mcdonalds-gives-its-restaurants-an-ai-makeover-2134f01e](https://wsj.com/articles/mcdonalds-gives-its-restaurants-an-ai-makeover-2134f01e)

**Digital Twins**

Digital twins are virtual representations of physical products, systems, or environments that use data from their real counterparts to reflect how they operate. They allow product teams to test scenarios, anticipate failures, and improve designs before making expensive changes in the physical world.

* **PepsiCo uses digital twins to trial plant changes** [*Supply Chain Dive*]
  [https://supplychaindive.com/news/pepsico-uses-digital-twins-to-trial-plant-changes-nvidia-siemens/809558/](https://supplychaindive.com/news/pepsico-uses-digital-twins-to-trial-plant-changes-nvidia-siemens/809558/)

* **Unilever builds more than 40 new digital twins** [*Manufacturing Digital*]
  [https://manufacturingdigital.com/articles/how-unilever-is-building-more-than-40-new-digital-twins](https://manufacturingdigital.com/articles/how-unilever-is-building-more-than-40-new-digital-twins)

* **Lowe’s introduces digital twins of its stores** [*Charlotte Business Journal*]
  [https://bizjournals.com/charlotte/news/2022/09/21/lowes-technology-charlotte.html](https://bizjournals.com/charlotte/news/2022/09/21/lowes-technology-charlotte.html)

**Blockchain** (Cryptography, Zero-Knowledge Proofs, Stablecoins)

Blockchain uses cryptography to maintain shared, tamper-resistant records, creating ways to coordinate transactions and ownership across participants. Related technologies include zero-knowledge proofs, which verify a claim without exposing the underlying private information, and stablecoins, digital tokens designed to track a reference value such as the dollar. For product leaders, these tools open opportunities in payments, identity, and verification, where their value depends on making trust and coordination easier for users.

* **Aura takes shape: luxury brands weigh the future of blockchain** [*Vogue Business*]
  [https://voguebusiness.com/story/technology/aura-takes-shape-prada-otb-and-lvmh-weigh-the-future-of-luxurys-blockchain-solution](https://voguebusiness.com/story/technology/aura-takes-shape-prada-otb-and-lvmh-weigh-the-future-of-luxurys-blockchain-solution)

* **Google Wallet expands digital IDs and privacy features** [*Lifewire*]
  [https://lifewire.com/google-wallet-new-features-11724766](https://lifewire.com/google-wallet-new-features-11724766)

* **JPMorgan rolls out JPM Coin deposit tokens on Base** [*The Block*]
  [https://theblock.co/news/business/2025-11-12-jpmorgan-deposit-token-jpm-coin-378493](https://theblock.co/news/business/2025-11-12-jpmorgan-deposit-token-jpm-coin-378493)

**Longevity**

Longevity focuses on extending healthy years of life, while HealthTech more broadly uses technology to improve prevention, diagnosis, treatment, and access to care. For product leaders, this creates opportunities to turn health data into useful interventions and everyday behaviors, with success measured through credible health outcomes and sustained use.

* **FDA authorizes hearing-aid software for AirPods Pro** [*Reuters*]
  [https://reuters.com/business/healthcare-pharmaceuticals/fda-authorizes-first-otc-hearing-aid-software-be-used-airpods-pro-headphones-2024-09-12/](https://reuters.com/business/healthcare-pharmaceuticals/fda-authorizes-first-otc-hearing-aid-software-be-used-airpods-pro-headphones-2024-09-12/)

* **L’Oréal unveils a device to analyze skin and ingredient suitability** [*The Verge*]
  [https://theverge.com/2025/1/6/24334941/loreal-cell-bioprint-ces-2025-skincare-beauty-tech](https://theverge.com/2025/1/6/24334941/loreal-cell-bioprint-ces-2025-skincare-beauty-tech)

* **Abbott launches an over-the-counter glucose monitor** [*Reuters*]
  [https://reuters.com/business/healthcare-pharmaceuticals/abbott-launches-over-the-counter-continuous-glucose-monitor-us-2024-09-05/](https://reuters.com/business/healthcare-pharmaceuticals/abbott-launches-over-the-counter-continuous-glucose-monitor-us-2024-09-05/)

**Metaverse**

The metaverse describes a vision of persistent, shared digital environments where people interact, work, play, and participate in virtual economies. For product leaders, it offers a space to experiment with digital identity, community, and ownership, while testing whether those experiences give people a compelling reason to return.

* **Manufacturing is keeping the industrial metaverse alive** [*WIRED*]
  [https://wired.com/story/the-metaverse-is-here-and-its-industrial/](https://wired.com/story/the-metaverse-is-here-and-its-industrial/)

* **IKEA hires people to work in its Roblox store** [*TIME*]
  [https://time.com/6985642/ikea-hiring-workers-roblox-game/](https://time.com/6985642/ikea-hiring-workers-roblox-game/)

* **LEGO Fortnite adds a social roleplaying city** [*The Verge*]
  [https://theverge.com/2024/12/9/24316954/lego-fortnite-brick-life-social-roleplay-game](https://theverge.com/2024/12/9/24316954/lego-fortnite-brick-life-social-roleplay-game)

**AR/VR**

Augmented reality (AR) overlays digital information onto the physical world, while virtual reality (VR) places users inside an immersive digital environment. These technologies let product teams explore spatial experiences, such as previewing a product in someone’s home or practicing a complex task, where interacting in three dimensions can improve understanding or performance.

* **Ray-Ban glasses gain an integrated display** [*The Guardian*]
  [https://theguardian.com/technology/2025/sep/17/meta-ray-ban-smart-glasses](https://theguardian.com/technology/2025/sep/17/meta-ray-ban-smart-glasses)

* **Amazon tests smart glasses for delivery drivers** [*Supply Chain Dive*]
  [https://supplychaindive.com/news/amazon-delivery-glasses-fulfillment-robots-ai-model/803748/](https://supplychaindive.com/news/amazon-delivery-glasses-fulfillment-robots-ai-model/803748/)

* **Live immersive Lakers games come to Vision Pro** [*The Verge*]
  [https://theverge.com/news/797970/apple-vision-pro-immersive-nba-lakers-games](https://theverge.com/news/797970/apple-vision-pro-immersive-nba-lakers-games)

**Robotics**

AI expands what robots can do by helping them interpret their surroundings, understand instructions, and adapt their actions to changing conditions. For product leaders, this brings intelligent software into the physical world, where usefulness depends on reliable performance under real operating conditions and safe interaction with people.

* **John Deere expands its autonomous machinery lineup** [*Reuters*]
  [https://reuters.com/technology/deere-boosts-bet-autonomous-tractors-with-new-machines-2025-01-06/](https://reuters.com/technology/deere-boosts-bet-autonomous-tractors-with-new-machines-2025-01-06/)

* **Mercedes invests in Apptronik and tests factory robots** [*Reuters*]
  [https://reuters.com/business/autos-transportation/mercedes-benz-takes-stake-robotics-maker-apptronik-tests-robots-factories-2025-03-18/](https://reuters.com/business/autos-transportation/mercedes-benz-takes-stake-robotics-maker-apptronik-tests-robots-factories-2025-03-18/)

* **Walmart and Symbotic deepen their robotics partnership** [*Reuters*]
  [https://reuters.com/markets/deals/automation-firm-symbotic-signs-200-mln-deal-buy-walmarts-robotics-unit-2025-01-16/](https://reuters.com/markets/deals/automation-firm-symbotic-signs-200-mln-deal-buy-walmarts-robotics-unit-2025-01-16/)

**Quantum Computing and Quantum Resilience**

Quantum computing uses quantum physics to perform computation, offering potential advantages for certain problems, such as simulating molecules and materials. Quantum resilience means preparing systems to withstand future quantum threats, particularly by adopting cryptography designed to resist quantum attacks. Product leaders should explore where quantum capabilities could eventually create value while planning how products will protect long-lived data and adapt their security over time.

* **Apple introduces post-quantum encryption for iMessage** [*WIRED*]
  [https://wired.com/story/apple-pq3-post-quantum-encryption](https://wired.com/story/apple-pq3-post-quantum-encryption)

* **IBM runs a quantum error-correction algorithm on conventional AMD chips** [*Reuters*]
  [https://reuters.com/business/ibm-says-key-quantum-computing-algorithm-can-run-conventional-amd-chips-2025-10-24/](https://reuters.com/business/ibm-says-key-quantum-computing-algorithm-can-run-conventional-amd-chips-2025-10-24/)

* **Google announces its Quantum Echoes algorithm** [*Reuters*]
  [https://reuters.com/technology/google-says-it-has-developed-landmark-quantum-computing-algorithm-2025-10-22/](https://reuters.com/technology/google-says-it-has-developed-landmark-quantum-computing-algorithm-2025-10-22/)

# Tools of the Trade

Overview: The software product teams actually live in. Tools change constantly and none of them will rescue a team that hasn't agreed on how it works — pick the ones that reduce the number of places the truth lives.

**Project / Product Management**

Connect strategy to execution by defining priorities, assigning ownership, and deciding what gets built next.

|  | Linear — [https://linear.app/](https://linear.app/)  Product planning, issues, development cycles, and coordination between people and agents. |
| :---- | :---- |
|  | **Jira** — [https://atlassian.com/software/jira/](https://www.atlassian.com/software/jira/) Backlogs, sprint planning, dependencies, and engineering delivery at scale. |
|  | **Asana** — [https://asana.com/](https://asana.com/)  Cross-functional projects, ownership, goals, and workflow automation. |
|  | **ClickUp** — [https://clickup.com/](https://clickup.com/)  Tasks, documents, dashboards, and AI assistance in one workspace. |
|  | **Productboard** — [https://www.productboard.com/](https://www.productboard.com/)  Connecting customer feedback to prioritization, product strategy, and roadmaps. |
|  | **ChatPRD** — [https://www.chatprd.ai/](https://www.chatprd.ai/)  Turning ideas and research into requirements, user stories, and product briefs.  |

**Research**

Understanding customers, markets, and actual behavior gives teams better evidence for where to place their next bet.

|  | Amplitude — [https://www.amplitude.com/](https://www.amplitude.com/)  Understanding adoption, funnels, retention, and customer behavior after release. |
| :---- | :---- |
|  | **UserTesting** — [https://www.usertesting.com/](https://www.usertesting.com/)  Recruiting participants and observing how people experience products. |
|  | **Dovetail** — [https://dovetail.com/](https://dovetail.com/)  Organizing customer feedback, using AI to surface themes across interviews and other sources. |
|  | **Maze** — [https://maze.co/](https://maze.co/)  Prototype testing, usability studies, surveys, and rapid customer feedback. |
|  | **Outset** — [https://www.outset.ai/](https://www.outset.ai/)  Customer understanding via context-aware AI-moderated interviews and simulated human data. |
|  | **Granola** — [https://www.granola.ai/](https://www.granola.ai/)  Capturing customer conversations and turning meeting notes into usable research inputs. |

**Design**

Turn product ideas into experiences people can see and interact with, making alternatives easier to explore and refine.

|  | Figma — [https://www.figma.com/](https://www.figma.com/)  Interface design, design systems, collaboration, and developer handoff. |
| :---- | :---- |
|  | **Webflow** — [https://webflow.com/](https://webflow.com/)  Building production websites visually, with content management and AI capabilities. |
|  | **Miro** — [https://miro.com/](https://miro.com/)  Journey maps, workshops, diagrams, and collaborative product exploration. |
|  | **Canva** — [https://www.canva.com/](https://www.canva.com/)  Fast visual communication, presentations, and campaign assets with AI-assisted creation. |
|  | **Adobe Firefly** — [https://adobe.com/products/firefly.html](https://www.adobe.com/products/firefly.html)  Generating and editing imagery and other creative assets for product concepts. |
|  | **Recraft** — [https://www.recraft.ai/](https://www.recraft.ai/)  Generating illustrations, icons, vectors, and consistent visual assets.  |

**Development**

Build, test, and maintain working software with tools that help engineers move faster and keep quality high.

|  | GitHub — [https://github.com/](https://github.com/)  Source control, pull requests, code review, and development automation. |
| :---- | :---- |
|  | **Cursor** — [https://cursor.com/](https://cursor.com/)  AI-assisted development with codebase context and agents that implement changes. |
|  | **Claude Code** — [https://claude.com/product/claude-code](https://claude.com/product/claude-code)  Delegating coding, debugging, refactoring, and testing tasks to an agent. |
|  | **OpenAI Codex** — [https://openai.com/codex/](https://openai.com/codex/)  Agent-assisted implementation, code review, debugging, and testing. |
|  | **Visual Studio Code** — [https://code.visualstudio.com/](https://code.visualstudio.com/)  Code editing, debugging, extensions, and integrated AI development workflows. |
|  | **Devin** — [https://devin.ai/](https://devin.ai/)  Delegating engineering work and managing local and cloud coding agents.  |

**AI: Rapid Prototyping & Testing**

Faster, more affordable prototypes give teams more shots on goal and help reveal which ideas deserve further investment.

|  | Lovable — [https://lovable.dev/](https://lovable.dev/)  Turning a product description into a working web application that can be shared and refined. |
| :---- | :---- |
|  | **Replit** — [https://replit.com/](https://replit.com/)  Creating, running, and deploying functional prototypes with an AI agent. |
|  | **v0** — [https://v0.app/](https://v0.app/)  Generating interfaces and working web applications, especially within the Vercel ecosystem. |
|  | **Bolt** — [https://bolt.new/](https://bolt.new/)  Building websites and application prototypes through prompts in the browser. |
|  | **Figma Make** — [https://www.figma.com/make/](https://www.figma.com/make/)  Turning prompts and designs into interactive, code-backed prototypes. |
|  | **Google AI Studio** — [https://aistudio.google.com/](https://aistudio.google.com/)  Prototyping AI-powered applications and trying model-driven interactions. |

# The Anatomy of a PRD

Introduction: The PRD is where a fuzzy idea becomes something a team can actually build. What follows is a generic template for product-initiated projects, aimed mostly at customer-facing teams; it isn't exhaustive and it won't fit every project, so adapt it per use as appropriate.

**The purpose of the PRD**

A PRD is a living guide written by a Product Manager used to communicate and outline what they are building, who it is for, and how it benefits the end-user. It also serves as a guide for business and technical teams to help develop, launch, and market the product.

A well-written PRD will help ensure that all parties involved in the development process are aligned and that the product meets the needs of the intended audience. Product Managers use this to drive stakeholder alignment, prioritize features, and avoid scope creep.

**The Foundation of the PRD:**

1. **Define the problem:** Clearly define the problem the product is intended to solve.

2. **Identify the audience:** Identify the intended audience for the product and their needs.

3. **Define the features:** Define the features and functionality required to meet the needs of the intended audience.

4. **Prioritize features:** Prioritize the features based on their importance and the resources available.

5. **Communicate clearly:** Use clear and concise language to communicate the requirements of the product.

6. **Get feedback:** Get feedback from stakeholders and iterate on the PRD as needed.

**Initial planning and forming internal alignment** — In the beginning of a project, the PRD typically serves as a place to brain-dump notes, ideas, and data as product, design, engineering teams align on what the problem is that  they are trying to solve and what the scope of the solution should be. During this initial phase, product managers are frequently conducting customer discovery to pinpoint who the user is and what the need is. Designers are starting to mock up early solution concepts while engineering teams begin to come up with possible technical solutions while validating technical feasibility. Together each team typically converges on a clear solution to pursue, and have a broad stroke understanding of scope. This phase traditionally concludes when the team has confidence the proposed solution will achieve the objective they are solving for, design mocks are polished, and we have a clear line of sight on how to build.

It is at this point that the PRD begins to be transformed into a digestible (living) document utilizing the following outline:

| Section | Description |
| :---- | :---- |
| **Title** | Include a descriptive title for the specific feature being developed. It should be clear that this document is a PRD. (i.e. Put “PRD” in the title). |
| **Authors** | Include a list of all of the main owners of the document. Readers can direct questions to authors; the expectation is that any questions/comments on PRD are addressed and resolved by the authors.  |
| **Reviewers / Approvers** | Include a list of important internal stakeholders to ensure proper visibility and approval while reviewing the PRD. Consider using a RACI matrix. |
| **Status** | Include an indication of the version or where in the lifecycle the PRD is currently at (e.g. “Draft”). The expectation is that the PRD is a living document and thus should always contain the most up-to-date version. |
| **Context / Problem Statement** | Include a high-level introduction to the project; this is typically formatted as a quick “elevator pitch” that briefly communicates the concept. It is recommended to justify the need with customer quotes, surveys, competitive research, and usage data. Additionally, discuss how the features fit in with the broader product context and align with the broader company strategy/vision. |
| **Objectives / Goals** | Clearly articulate who the customer is and what is trying to be achieved for whom with this product/feature. Typically this is a set of bullet points that describe high-level goals, ideally in priority order. |
| **Non-Goals** | In the interest of all-around clarity, it is recommended to always take the time to outline non-goals of the project. Utilize this opportunity to include potential ancillary benefits that are NOT going to materialize as a result of this project and make sure that they are appropriately covered. |
| **KPIs / Success Metrics** | Include specific, quantitative, metrics/KPIs that are expected to move and by how much as a result of this project being successfully accomplished. This section can be challenging to write but critically important.  |
| **Use Cases / Narrative** | Include more context around the primary customer for this project and what the main use cases look like — help “paint the picture” of how this product/feature will be used. It is recommended to Include user personas and user scenarios. |
| **Product Requirements** | Include a thorough list of what is in scope and what is out of scope — including  user experience requirements *(external)*, errors and edge cases *(external)* as well as reporting requirements *(Internal)*. It is recommended to use the MSCW Method to capture more clearly all features in and out.  |
| **Timeline / Rollout plan** | Include a high-level timeline for major release milestones along the project’s journey. This is an opportunity to consider at what stage the project is at and how it will graduate between GA, internal testing, usability study, pilot/alpha, beta, staged rollout, feature flag, locale, etc. Milestones and timelines are not required to be tracked in the PRD necessarily — however, it is useful to link it to the document where the development progress is being tracked. |
| **Dependencies / Risks** | Include an overview of what could go wrong with this product, specifically what mitigations need to exist to manage these risks as well as any cross functional dependencies that the project has. |
| **Q&A / Open Issues** | Include a list of what details are yet to be resolved as well as an area to cover common questions and answers surrounding the project.  |
| **Other Considerations** | Include, typically as an Appendix, any additional legal considerations, support considerations and content, tax considerations, or comms considerations. |

**Continue Down the Rabbit Hole …**

* How to Write a PRD That Actually Helps You Build Products [*Reforge*] [https://www.reforge.com/blog/evolving-product-requirement-documents](https://www.reforge.com/blog/evolving-product-requirement-documents)

* How to Write a Painless Product Requirements Document [*Jerry Cao*] [https://uxpin.medium.com/how-to-write-a-painless-product-requirements-document-508ff6807b4a](https://uxpin.medium.com/how-to-write-a-painless-product-requirements-document-508ff6807b4a)

* PRD template from the former head of product at Square: [https://docs.google.com/document/d/1mEMDcHmtQ6twzNlpvF-9maNlAcezpWDtCnyIqWkODZs/](https://docs.google.com/document/d/1mEMDcHmtQ6twzNlpvF-9maNlAcezpWDtCnyIqWkODZs/edit#heading=h.7ueoyja6ijay)

* PRD template from the team at Figma
  [https://coda.io/@yuhki/figmas-approach-to-product-requirement-docs/prd-name-of-project-1](https://coda.io/@yuhki/figmas-approach-to-product-requirement-docs/prd-name-of-project-1)

* PRD Template from Product Hunt
  [https://docs.google.com/document/d/1yrU5F6Gxhkfma91wf_IbZfexw8_fahbGQLW3EvwdfQI/](https://docs.google.com/document/d/1yrU5F6Gxhkfma91wf_IbZfexw8_fahbGQLW3EvwdfQI/edit#heading=h.rmt3yiu4sfqd)

* Bolt PRD Template (Ideal for hardware products)
  [https://docs.google.com/document/d/1eBaBOF2A3rIfXeqqjq4uFAxX2f85JqanJmq-72ptur0/](https://docs.google.com/document/d/1eBaBOF2A3rIfXeqqjq4uFAxX2f85JqanJmq-72ptur0/)

Product Requirements Document (PRD) Template by *ProductSchool*  A Product Requirements Document template from Product School for defining your product’s purpose, users, features, and functionality, helping your team align on what to build and why.[https://productschool.com/resources/templates/prd](https://productschool.com/resources/templates/prd)

#

# Acronym Cheat Sheet (A.C.S.)

Overview: A curated glossary of the most essential Product Management acronyms. These are just a starting point — there are hundreds more that you’ll learn along the way.

| Acronym | Stands For | Definition |
| :---: | :---- | :---- |
| **A/B** | A/B Testing | An experimental method comparing two versions (A and B) to see which performs better. Commonly used for optimizing UX and marketing. |
| **ACV** | Annual Contract Value | The annualized revenue per customer contract, often used for subscriptions. Useful for sizing deals and forecasting. |
| **AI** | Artificial Intelligence | The simulation of human intelligence in machines that can learn, reason, and make decisions. Powers modern automation. |
| **AP** | Accounts Payable | Money a company owes to suppliers for goods and services received. Managed as a short-term liability and key to cash flow. |
| **API** | Application Programming Interface | A set of rules and tools that allow software systems to communicate. Critical for integrations and platform strategies. |
| **AR** | Accounts Receivable | Money owed to a company by customers for delivered goods or services. Reflects pending inflows and short-term liquidity. |
| **ARR** | Annual Recurring Revenue | The yearly equivalent of MRR. Used for forecasting and valuation in recurring revenue businesses. |
| **B2B** | Business to Business | Companies selling products or services to other businesses. Often characterized by longer sales cycles and contracts. |
| **B2C** | Business to Consumer | Companies selling products or services to individual consumers rather than businesses. Usually higher volume, shorter sales cycles, and broader marketing. |
| **CAC** | Customer Acquisition Cost | The total cost to acquire one new customer. A key metric for marketing efficiency and scalability. |
| **CLV** | Customer Lifetime Value | Another term for LTV; emphasizes long-term relationship profitability. Often compared to CAC for unit economics. |
| **COGS** | Cost of Goods Sold | The direct costs of producing goods or services sold by a company. A key component of gross margin analysis. |
| **CPC** | Cost Per Click | The amount paid each time a user clicks an ad. Core metric for performance-based digital advertising and user acquisition.  |
| **CPM** | Cost Per Mille | Advertising metric representing cost per thousand impressions. Used to evaluate brand awareness and reach efficiency. |
| **DTC** | Direct to Consumer | Brands selling straight to customers, bypassing retailers and other intermediaries.  |
| **EBITDA** | Earnings Before Interest, Taxes, Depreciation, and Amortization | A measure of operational profitability before accounting for financing and non-cash expenses. Used for valuation comparisons. |
| **GM** | Gross Margin | The percentage of revenue left after subtracting COGS. Reflects pricing power and efficiency. |
| **GMV** | Gross Merchandise Value | The total value of goods sold through a marketplace platform. Commonly used by e-commerce and fintech companies. |
| **GPT** | Generative Pretrained Transformer | A large language model architecture trained on vast text corpora to generate and understand language. Powers modern AI apps. |
| **GTM** | Go-To-Market  | A plan for launching and scaling a product in its target market. Aligns marketing, sales, and product. |
| **IA** | Information Architecture | The organization and structure of information within a digital product. Helps users find what they need intuitively. |
| **IaaS** | Infrastructure as a Service | Cloud-based computing infrastructure provided on demand. Examples include AWS and Google Cloud. |
| **ICP** | Ideal Customer Profile | The description of the customer most likely to benefit from your solution. Guides sales and marketing targeting. |
| **IPO** | Initial Public Offering | The first time a company’s shares are offered to the public. Marks the transition from private to public ownership. |
| **JTBD** | Jobs To Be Done | A framework for understanding why users “hire” a product to accomplish specific outcomes. Helps define product-market fit through user motivations. |
| **KPI** | Key Performance Indicator | A measurable metric that indicates progress toward strategic goals. Used to track success in product, marketing, and operations. |
| **LLM** | Large Language Model | A neural network trained on vast text datasets to understand and generate human-like language. Examples include GPT models. |
| **LTV** | Lifetime Value | The total revenue expected from a customer over their relationship with a business. Helps evaluate sustainable growth. |
| **MCP** | Model Context Protocol | An open standard for connecting AI models to tools and data sources. Simplifies tool discovery, invocation, and secure context sharing. |
| **ML** | Machine Learning | A subset of AI where algorithms improve automatically through data and experience. Drives prediction and personalization. |
| **MoM** | Month-over-Month | A metric showing short-term growth or decline. Useful for tracking performance trends. |
| **MRR** | Monthly Recurring Revenue | The predictable revenue generated every month from subscriptions. Key indicator of SaaS business health. |
| **MSA** | Master Services Agreement | A contract setting overarching legal and commercial terms for a client–vendor relationship. Individual projects use add-on orders. |
| **MVP** | Minimum Viable Product | The simplest version of a product that can be released to test assumptions and gather user feedback. Focuses on learning quickly, not perfection. |
| **NDA** | Non-Disclosure Agreement | A legal contract protecting confidential information. Common in early product collaboration. |
| **NLP** | Natural Language Processing | The AI field enabling computers to interpret and generate human language. Used in chatbots and voice assistants. |
| **NPS** | Net Promoter Score | A measure of customer loyalty based on likelihood to recommend. Ranges from detractors to promoters. |
| **NPV** | Net Present Value | The present value of future cash flows minus the initial investment. Positive NPV indicates a good investment. |
| **OKR** | Objectives and Key Results | A goal-setting framework aligning teams around measurable outcomes. Encourages focus, transparency, and accountability. |
| **P&L** | Profit and Loss Statement | A financial statement summarizing revenues, costs, and expenses over time. Shows a company’s ability to generate profit. |
| **PaaS** | Platform as a Service | A cloud computing model that offers developers a platform to build, test, and deploy applications. Sits between IaaS and SaaS. |
| **PLG** | Product-Led Growth | A go-to-market strategy where the product drives acquisition, conversion, and retention. Often paired with freemium models. |
| **PM** | Product Manager | The role responsible for defining product vision, prioritization, and execution. The “CEO of the product.” |
| **PMF** | Product-Market Fit | The degree to which a product satisfies a strong market demand. Often measured by retention, referrals, and rapid growth. |
| **PRD** | Product Requirements Document | A document outlining what a product should do and why. It serves as a bridge between strategy and development. |
| **RFP** | Request for Proposal | A document inviting vendors to propose solutions to a business problem. Used in procurement and partnerships. |
| **ROI** | Return on Investment | A metric measuring profitability relative to cost. Used to evaluate the efficiency of an investment. |
| **SaaS** | Software as a Service | A software delivery model where users access applications via the internet. Offers recurring revenue and scalability. |
| **SAM** | Serviceable Available Market | The portion of TAM targeted by your products within your reach. Helps refine focus and forecast realistic growth. |
| **SDK** | Software Development Kit | A collection of tools, code, and documentation to build software on specific platforms. Used to extend product capabilities. |
| **SLA** | Service Level Agreement | A contractual commitment defining performance standards and uptime targets. Common in SaaS and vendor relationships. |
| **SLG** | Sales-Led Growth | A go-to-market motion where sales drives acquisition and expansion, with product and marketing supporting lead generation. |
| **SOM** | Serviceable Obtainable Market | The segment of the SAM that your business can realistically capture. Often used in investor decks. |
| **SOW** | Statement of Work | A project-specific document under an MSA that defines scope, deliverables, timelines, and pricing. Used to control execution. |
| **SQL** | Structured Query Language | A language used to manage and query relational databases. Essential for data analysis and backend systems. |
| **SWOT** | Strengths, Weaknesses, Opportunities, Threats | A strategic tool for assessing internal and external factors that influence success. Common in product planning and market entry. |
| **TAM** | Total Addressable Market | The total revenue opportunity available for a product or service. Often the first sizing step in market analysis. |
| **TCO** | Total Cost of Ownership | The full lifecycle cost of an asset or system, including acquisition, operation, and maintenance. Guides long-term ROI decisions. |
| **TCV** | Total Contract Value | The full value of a contract over its entire term, including recurring fees and one-time charges. Guides pipeline and valuation. |
| **VC** | Venture Capital | Financing provided to startups with high growth potential. Typically in exchange for equity. |
| **YoY** | Year-over-Year | A comparison of performance between the same period in two years. Smooths out seasonality. |

# Book Club [Recommended Reading]

Overview: A short shelf, chosen for range rather than completeness. Read three of these properly and you'll carry more useful frameworks than you would from skimming all of them.

|  | Crossing the Chasm by Geoffrey A. Moore Moore's account of the gap between early adopters and the mainstream, and what a product has to change to cross it. |
| :---- | :---- |
|  | **The Power Law** by Sebastian Mallaby A history of venture capital and the bets that defined it. Useful for how investors think about a portfolio: most positions return little, a few return everything, and the discipline is funding the second kind long enough to find out. |
|  | **Thinking in Bets** by Annie Duke Duke, a former poker champion, on separating the quality of a decision from the quality of its outcome. Directly useful when you review what worked and what didn't on your own projects. |
|  | **The Lean Startup** by Eric Ries The book that made build-measure-learn standard vocabulary. Parts of it now read as dated, but it remains the clearest argument for treating a launch as an experiment rather than a finish line. |
|  | **Zero to One** by Peter Thiel Thiel argues that competition is overrated and that the valuable companies are the ones doing something nobody else is doing. You don't have to accept every conclusion to find his questions about differentiation worth answering. It reads more as a manifesto than a manual. |
|  | **The Innovator's Dilemma** by Clayton M. Christensen Christensen's explanation of why competent companies lose to worse products: the decisions that protect a profitable business are the same ones that prevent it from pursuing a smaller, less attractive market until someone else owns it. Pairs with the Barbell chapter. |
|  | **Measure What Matters** by John Doerr Doerr's case for OKRs, drawn from Intel and Google. The examples are more useful than the evangelism: watch how teams write a measurable key result instead of restating an ambition as a goal. |
|  | **Algorithms to Live By** by Brian Christian and Tom Griffiths Christian and Griffiths on what computer science says about everyday decisions: when to stop looking, how to schedule, how much to explore before committing to one option. |
|  | **That Will Never Work** by Marc Randolph Randolph's account of Netflix's first years, including the ideas that failed. A good antidote to founding stories that make the outcome sound inevitable. |
|  | **David and Goliath** by Malcolm Gladwell Gladwell on how an advantage can become a liability and a disadvantage can become an edge. Treat the stories as prompts for your own strategy, not as evidence. |
|  | **Build** by Tony Fadell Fadell worked on the iPod, the iPhone, and the Nest thermostat, and writes about all of it with unusual specificity. The most practical book here on the day-to-day work of leading teams and the decisions nobody usually writes down. |

**MPD-400:** Master of Product Design and Development Management    /    **mpd²**