Appendix

Appendix: Templates

These templates are starting points. Copy them, adapt them to your product and your organization, and keep them short. A template that nobody fills in is worse than none.

A. Problem statement

[Who] need to [goal or job], but [obstacle], which leads to [consequence, ideally measurable]. Today they [current workaround], which falls short because [gap].

Check it:

See Chapter 2.

B. One-page product strategy

Vision. One sentence describing the world you want to create for customers.

Diagnosis. What is really going on in the market, and the hardest problem you face. Three to five sentences.

Where we play. The customers, problems and channels you will focus on.

How we win. The advantage that will make those customers choose you, and why competitors with the same AI tools cannot copy it in a quarter.

Guiding policy. Your overall approach, in one or two sentences.

Key actions. Three to five coherent actions for the next year.

Outcomes. The North Star and the input metrics you will use to know it is working.

Not doing. What you will stop or decline, with a sentence on why.

See Chapter 4.

C. Product brief

Title and owner. The name of the work and the product trio accountable for it.

Problem. What problem, for whom, with what evidence. Link the problem statement.

Outcome measure. The measure you expect to move, today’s baseline, and the target.

Who it is for. Users, buyers and beneficiaries.

What we know. Evidence from discovery.

What we assume. The riskiest assumptions, and how and when you will test them.

Scope. What is in the first slice.

Out of scope. What is explicitly not included, and why.

Must include. Non-negotiable requirements: privacy, security, audit, accessibility, performance.

Open questions. What the team will answer as it builds.

Decision log. Dated decisions, with the reasoning.

See Chapter 9.

D. One-page roadmap

Now (committed) Next (likely) Later (exploring)
Problem or outcome, with its measure Problem, with its measure Problem only
… … …
Dated commitments, marked clearly

Capacity allocation this quarter: new outcomes %, reliability and performance %, security and compliance %, technical debt %, discovery and experiments __%.

See Chapter 5.

E. Business case

  1. Problem and customers: from discovery, with evidence.
  2. Revenue effect: new customers, expansion, churn prevented, price supported. Show a low, likely and high case, with assumptions.
  3. Cost to build: product, design and engineering time; third-party costs.
  4. Cost to run: hosting, support, maintenance, AI model costs, partner fees, per customer and in total.
  5. Effect on unit economics: gross margin, payback period, retention.
  6. Risks: what has to be true, and what could go wrong.
  7. How we will know: the measure, the threshold and the date you will check.

See Chapter 6.

F. Pilot agreement

  • Problem and measure: the outcome being tested and how it will be measured.
  • Baseline: the measure today.
  • Success threshold: the result that justifies a purchase.
  • Scope and duration: teams, sites, users and dates. Usually 60 to 90 days.
  • Responsibilities: what each side provides, including data, people and time.
  • Price: the pilot fee.
  • Decision-maker and date: who decides, and when.
  • Expansion terms: the price, scope and timeline that apply if the threshold is met.

See Chapters 3 and 13.

G. Launch readiness checklist

Product

Quality, security and trust

Customers

Go-to-market

Support and operations

After launch

See Chapters 9 and 13.

H. Customer health score

Signal Example measures Weight
Outcomes Customer’s own outcome measure, trended Highest
Adoption Active users by site; usage trend High
Relationship Executive sponsor engaged; champion in place Medium
Support Open issues; ticket trend Medium
Commercial Payment timeliness; contract questions Low

Every at-risk account has an owner and a written plan. See Chapter 14.

I. AI feature evaluation plan

  • Feature and purpose: what the AI does, for whom.
  • Autonomy level: assist, suggest, act with approval, act within limits, or act autonomously, and why.
  • Cost of an error: what happens if the output is wrong and nobody notices.
  • Evaluation set: source, size, and how hard and unusual cases are included; how sensitive data is removed.
  • Quality criteria: specific, agreed by experts.
  • Scoring method: expert review, automated checks, AI-assisted grading calibrated against people.
  • Launch thresholds: what quality is required to launch; what blocks a release.
  • Fairness checks: groups compared, and acceptable differences.
  • Production monitoring: acceptance and edit rates, error reports, sampled review, cost and latency.
  • Owner: the person accountable for quality.

See Chapter 17.

J. AI agent job description

  • Name and purpose: what the agent does, in one sentence.
  • Owner: the person accountable for it.
  • Inputs: the data and sources it may use.
  • Outputs: what it produces, and for whom.
  • Permissions: the systems and actions it may access, and nothing more.
  • Human checkpoints: which outputs or actions require review or approval.
  • Quality measures: how its work is checked, and how often.
  • Escalation: what it does when it is unsure, and how people raise concerns about it.
  • Review date: when its job description and permissions will next be reviewed.

See Chapters 8 and 17.

K. Glossary

Acceptance criteria. The specific, testable conditions a user story must meet to be considered done.

Annual recurring revenue (ARR). The yearly value of active subscriptions, excluding one-time fees.

Autonomy level. How much an AI feature or agent does without a person’s involvement: assist, suggest, act with approval, act within limits or act autonomously.

Business associate agreement (BAA). Under HIPAA, the contract required when a vendor handles protected health information on behalf of a healthcare provider or health plan.

Buying group. Everyone involved in a business purchase decision: users, champions, economic buyers, technical gatekeepers and influencers.

Cohort. A group of customers or users who started in the same period, tracked together over time.

Configuration. A setting that changes how the product behaves for a customer without new code.

Customer acquisition cost (CAC). Sales and marketing spending in a period, divided by the number of new customers won.

Discovery. The continuous work of learning what to build: understanding problems and testing solutions before and during delivery.

Error budget. The amount of unreliability a service is allowed under its service level objective. When it runs low, effort shifts to reliability.

Evaluation set. A collection of realistic examples, with agreed quality criteria, used to measure an AI feature’s output before release and after every change.

Feature flag. A switch that turns a capability on or off for specific customers or users without a new release.

Grounding. Giving an AI model the specific, relevant information it should use to answer, such as a customer’s records, rather than relying on its general knowledge.

Gross margin. Revenue minus the cost of delivering the service, as a percentage of revenue.

Gross revenue retention. The share of last year’s recurring revenue kept from existing customers, ignoring upsells.

Input metric. A measure a team can move directly, which drives the North Star.

Jobs to be done. The progress a customer is trying to make in a particular circumstance, which they “hire” a product to help with.

Lifetime value (LTV). The gross profit a customer is expected to generate over the whole relationship.

Minimum viable product (MVP). The smallest thing that delivers real value to real users and tests the core assumption.

Net revenue retention (NRR). The share of last year’s recurring revenue kept from existing customers, including expansion. Above 100 percent means existing customers are growing.

North Star metric. The single measure that best captures the value a product delivers to customers and leads long-term business results.

Outcome. A measurable change in customer behavior or results, as distinct from output, the features a team ships.

Payback period. How long it takes for a customer’s gross profit to repay the cost of acquiring them.

Positioning. How you want the market to understand what your product is, who it is for and why it is better than the alternatives.

Pricing metric. The unit a customer pays for, such as a seat, a transaction or a patient covered.

Product trio. The product manager, product designer and engineering lead who share responsibility for discovery and the outcome.

Prompt injection. An attack in which instructions hidden in content an AI system reads attempt to change its behavior.

Psychological safety. A shared belief that team members can take risks, ask questions and admit mistakes without fear of punishment or humiliation.

Retrieval-augmented generation. A common grounding method: finding relevant information and supplying it to a model with each request.

Service level objective (SLO). An explicit target for a service’s reliability or speed.

Smoke test. An experiment that presents a product or feature as available, to measure whether people try to use it before it is built.

Time to first value. How long it takes a new customer to see the outcome they bought the product for.

Unit economics. The revenue and costs associated with a single customer, used to judge whether growth makes the business better or worse.

Vanity metric. A number that looks good and goes up, but does not help you decide anything.

WCAG. The Web Content Accessibility Guidelines, the international standard for accessible digital products; Level AA is the common target.

Wizard of Oz test. An experiment in which people deliver a product’s outcome by hand behind something that looks like real software.

L. Further reading

These are the books and studies cited in this playbook that repay reading in full, grouped by the part of the job they help with.

The product leader’s job

  • Marty Cagan, Inspired: How to Create Tech Products Customers Love (2nd ed., 2017). The clearest description of how strong product teams work, and the source of the four big risks.
  • Teresa Torres, Continuous Discovery Habits (2021). A practical, weekly method for discovery, including the opportunity solution tree and the product trio.

Finding and validating problems

  • Rob Fitzpatrick, The Mom Test (2013). A short, funny and indispensable guide to customer interviews that produce the truth.
  • Clayton Christensen and colleagues, Competing Against Luck (2016). The jobs-to-be-done idea, explained by its best-known advocate.

Strategy, positioning and roadmaps

  • Richard Rumelt, Good Strategy Bad Strategy (2011). Diagnosis, guiding policy and coherent action, and why most “strategies” are nothing of the kind.
  • A.G. Lafley and Roger Martin, Playing to Win (2013). Where to play and how to win, from a company that practiced it.
  • April Dunford, Obviously Awesome (2019). The most practical method for positioning a product.
  • Joshua Seiden, Outcomes Over Output (2019). A short book that changes how teams plan.

The numbers

  • Michael Marn and Robert Rosiello, “Managing Price, Gaining Profit,” Harvard Business Review (1992). The classic case for taking pricing seriously.
  • Mehrdad Baghai, Stephen Coley and David White, The Alchemy of Growth (1999). The three horizons of growth.

Teams and delivery

  • Amy Edmondson, The Fearless Organization (2018). Psychological safety, from the researcher who defined it.
  • Nicole Forsgren, Jez Humble and Gene Kim, Accelerate (2018). The evidence that delivery speed and stability go together.
  • Donald Reinertsen, The Principles of Product Development Flow (2009). Cost of delay, batch size and queues, for readers who want the underlying economics.

Customers and design

  • Christine Sinsky and colleagues, “Allocation of Physician Time in Ambulatory Practice,” Annals of Internal Medicine (2016). A sobering picture of the work many healthcare products are meant to support.
  • Ann Cavoukian, Privacy by Design: The 7 Foundational Principles (2009). Short, clear and still the foundation of privacy-respecting products.

AI

  • National Institute of Standards and Technology, AI Risk Management Framework (2023). A structured, practical way to govern AI risk.
  • McKinsey & Company, The State of AI (published yearly). A useful annual snapshot of how organizations are adopting AI and agents.