Companion to The Product Manager’s New Job Is Acceptance

In this section

• Write One Page That Says Who Is Being Tested

  A test means something only for a defined population. This one page names it, its actors, what you know about it, and what you are still betting on.

• Draw Who Buys, Who Uses, and Who Can Force the Two Together

  A one-page map of every person who buys, approves, evaluates, administers, uses, or is affected by the product, and what each can do to the others.

• Name a Person and a Prediction at Every Stage of the Path

  A canvas for writing the stages an ICP passes through, with an actor, an expected action, evidence, and a likely failure at each one.

• State Each Hypothesis So It Can Come Out Wrong

  A fixed sentence shape for an acceptance hypothesis, with weak versions rewritten to be testable.

• Write Down What Counts as Failure Before You Run the Test

  One page per test, so someone else can run it under the same conditions and nobody can move the goalposts afterward.

• Record Every Time a Person Steps In to Make the Path Work

  A running log of every human intervention in a demo, trial, onboarding, or early use, with a category and a question about whether it could be built in.

• Keep One Page per Test That Informs a Decision So the Next Person Can Use It

  The sixteen fields of an Acceptance Evidence Record, with a completed example and a guide to confidence levels.

• Record the Decision, the Evidence Behind It, and the Dissent

  A log entry for each decision that matters, with the evidence used, the evidence missing, and what would make you decide again.

• Check Every Piece of Evidence Against the Fourteen Ways It Can Mislead

  A checklist of the fourteen failure modes from the chapter on metrics, each with one question to ask of a piece of evidence.

• Watch, Write It Down, and Do Not Help

  A short guide for anyone who observes a test: what to say, what not to say, what to record, and when to step in.

• Ask About Specific Behavior, Not About Opinions

  Questions that tie an interview to something you observed, so the customer explains behavior and does not predict or flatter.

• Log the Help Flag and the Actor with Every Event on the Path

  A sample of the fields an acceptance event should carry, including who intervened and whether use was voluntary.

• A Five-Minute Claim Fails at Setup, and a Four-Practice Test Shows It

  An illustrative intake product taken through the whole method: specification, promise audit, hypothesis, adversarial test, rescue ledger, record, and decision.

• A Hundred Percent Login Rate Does Not Show That Clinicians Accept the Tool

  An illustrative documentation tool bought by an executive and used by required clinicians, and how to look past the login count.

• A Tool a Professional Chooses Should Be Tested Against the Spreadsheet

  An illustrative workpaper tool for independent accountants, where the accountant chooses and pays, and repeated voluntary use is the evidence.

• Give Your AI Workforce These Prompts and Check What Comes Back

  Fifteen prompts for drafting, challenging, and updating acceptance work, each with the check to run on the output.

• The Standards and Studies Behind the Book, With What Each Can and Cannot Show

  Where to read further, with the limits of each source stated. Check current editions and dates before you quote.

• The 95 Percent AI Pilot Failure Figure and What Its Source Supports

  The claim that 95 percent of AI pilots fail is quoted everywhere. This page says what the source reported and what it did not, and it is kept current.