phase 1 · understand
Brief intake
Go deeper
The ICP is only as good as the brief under it. This page lists every field the brief needs, where each one comes from, how much to trust it, and which line of the ICP it feeds.
Every ICP line traces back.
Each "fits if" or "not if" line names the brief fields it came from. A line with no source gets cut.
Payers beat opinions.
What paying customers did and said outranks what the founder believes, which outranks what the model infers from the homepage.
Enough to start = every "must".
With no paying customers yet, the market fields become required and the whole ICP is labelled a guess until real payers confirm it.
how much to trust a field
Four levels of evidence
Every captured field carries one of these. When two sources disagree, the higher one wins.
A — Paying customers · What payers did and said: Stripe records, their answer to "why did you buy", cancel reasons.
B — How signups behave · Who signs up, who reaches the first result, who goes quiet. Real behavior, but nobody paid yet.
C — The market · Competitors' customers, public reviews, what buyers say in their communities.
D — Claims and guesses · The homepage, the founder's beliefs, anything the model infers.
the first field
Product stage decides what's possible
Read it before anything else. A waitlist can't offer a trial, a product with no signups has no warm moments, and only paying customers give tier A evidence. The Playbook Library's "pick your stage" step uses four of these: launching soon is waitlist, launched with nobody paying is live with no signups, signups with no payments matches the same stage here, and a few customers covers paying customers and growing.
| stage | how to tell from outside | the offer can be | ICP evidence available | what it unlocks |
|---|---|---|---|---|
| Pre-product | no page, or a page with nothing to sign up for | nothing yet | founder belief only | out of scope: the brief needs a URL with something on it |
| Waitlist | email capture, "coming soon", "join the waitlist" | join the waitlist, early access | market research, waitlist answers | launch week, owned content, community answers. No trials, no lifecycle plays |
| Live, no signups yet | signup or checkout button, pricing page, little public mention | a trial or free plan | market research, founder belief | founder-led outbound, community answers, offer posts |
| Live, signups but no payments | live product; founder says people sign up but don't pay | help getting set up, then the trial | adds signup behavior (tier B) | activation nudge, stalled-trial rescue, bringing back quiet signups |
| Live, paying customers | Stripe data, testimonials with names, "customers" page | the paid plan, with proof | adds payers (tier A): the ICP can copy them | share a new customer, comparison pages, win-back, failed payments |
| Growing | a channel that keeps producing customers | upgrades, more seats or products | enough payers to split best from worst | second-product expansion, referral badge; more of what paid |
The brief should store the full stage, its evidence, and the date it was last checked.
the checklist
Every field the brief should hold
Grouped by what it describes. The right-hand column says which part of the ICP (or the pipeline after it) the field feeds.
The market fields come from competitor research. The full method, with the channel map, is on the Market Research page.
The product what it is and what it takes to use it
| field | where it comes from | evidence | feeds | required |
|---|---|---|---|---|
| Product stage | homepage: waitlist form or signup and checkout, pricing page, "coming soon"; Stripe; interview | A · D | which playbooks can run, what the offer can be, and whether the ICP can lean on payers | must |
| Category, in the buyer's words | homepage headline; how reviewers describe it | D · C | search terms, and where to look for buyers | must |
| The job it does | homepage; onboarding | D | the pain words detectors match | must |
| What it replaces | homepage; interview; payers' "what did you use before" | A · D | which moments apply: doing it by hand, unhappy with a tool | must |
| What they must already have | integrations page; docs | D | hard filter: tools the buyer uses (e.g. QuickBooks, Shopify, Stripe) | must |
| The first result, and how long it takes | product; signup data | B | the activation definition; speed to value when ranking moments | must |
Price and buying who pays and how the decision is made
| field | where it comes from | evidence | feeds | required |
|---|---|---|---|---|
| Price and plans | pricing page; Stripe | A · D | budget, and the size of business that can afford it | must |
| Who uses it vs who pays | Stripe billing contact vs user accounts; interview | A · D | the role the filter targets | must |
| How buying happens | checkout vs 'book a demo' | D | in or out: buying committees are out of scope | must |
| Trial or free plan | pricing page | D | the offer a campaign can make | helps |
Who pays today the strongest evidence, when it exists
| field | where it comes from | evidence | feeds | required |
|---|---|---|---|---|
| Paying customers | Stripe: customer, domain, plan, start date | A | the traits to copy: size, industry, tools, location | must if payers |
| Why they bought, in their words | onboarding answer; emails; calls | A | which moments are real, and the opening lines | must if payers |
| How they heard about you | a signup question | A | channels that already work | must if payers |
| Best vs worst customers | retention, upgrades, support load | A · B | narrows the filter toward the best ones | helps |
| Why people cancelled | cancel reason | A | the "not if" list | must if payers |
| Who signed up and never paid | app data | B | traits that predict not paying | helps |
What the founder believes claims to test, not facts
| field | where it comes from | evidence | feeds | required |
|---|---|---|---|---|
| Who it's for | interview | D | the starting hypothesis | must |
| Who you'd turn away | interview | D | the "not if" list | must |
| Who asked for it first | interview | C · D | where real demand started | helps |
Market research outside evidence, required when nobody pays yet
| field | where it comes from | evidence | feeds | required |
|---|---|---|---|---|
| Competitive analysis | homepage; search; review sites | C | the competitive landscape; where unhappy customers are; comparison pages | must |
| Competitors' customers | reviews, case studies, logo walls | C | the filter's traits when there are no payers | must if payers |
| Watering holes | communities, forums, groups where buyers gather (Sales Safari) | C | detector sources | must |
| Voice of customer | Sales Safari passes: pain, jargon, worldview, recommendations | C | phrases for detectors and opening lines | must |
Limits where the product can't win yet
| field | where it comes from | evidence | feeds | required |
|---|---|---|---|---|
| Where it works | geography, language, compliance | D | hard filter | helps |
| When it works | seasons, deadlines, budget cycles | C · D | campaign timing | helps |
| What you can deliver now | interview | D | the whole-product gate on moments | must |
| Channels you already have | X audience, LinkedIn, email list, communities you're in | A | the reachable-through-a-channel gate | must |
the founder interview
Nine questions that fill the gaps
The homepage can't answer these. Ask them once, in the founder's words, and store the answers verbatim.
question 1 — Who paid you first, and what were they doing right before? · Fills: Who pays today · why they bought (tier A if it's about real payers) · → the starting moment
question 2 — What did they use before you? · Fills: What it replaces (tier A from payers, D otherwise) · → doing it by hand, unhappy with a tool
question 3 — Who would you refuse as a customer, and why? · Fills: Who you'd turn away (tier D) · → the "not if" list
question 4 — What has to be true for someone to get value on day one? · Fills: What they must already have · first result (tier D, confirmed by B) · → hard filters, activation
question 5 — Who uses it, and who pays for it? Same person? · Fills: Who uses it vs who pays (tier D) · → the role in the filter
question 6 — Where do the people you want spend time online? · Fills: Watering holes (tier D, confirmed by C) · → detector sources
question 7 — What do they call the problem, in their own words? · Fills: Voice of customer (tier D, confirmed by C) · → detector phrases, opening lines
question 8 — What channels do you already have, and how big are they? · Fills: Channels you already have (tier A) · → the reachability gate
question 9 — What do buyers expect that the product can't do yet? · Fills: What you can deliver now (tier D) · → the whole-product gate
assembling the icp
Turning fields into a filter
Observable
Each line must be checkable on a stranger from outside: team size, tools in use, a public price, where they posted. "Values speed" doesn't qualify.
Traced
Each line lists the fields it came from. Its confidence is the highest evidence tier among them.
Tested
The finished filter passes three tests: you can list them, count them, and they'd vouch for you to each other.
An example: a founder-tools product's "you fit if" lines, with where each one comes from and how sure each one is.
Who it's for — Paying customers
→
fits if: built alone or with a team you can count on one hand
founder belief, backed by a small number of payers
What they must already have — How buying happens
→
fits if: has a public price
a product rule: there has to be something to buy
Watering holes — Voice of customer
→
fits if: launched in the last 90 days, or about to
market evidence: the day 3 to 45 window after a launch
Who it's for
→
fits if: between $0 and a few hundred in MRR, and knows the number
founder belief only; test against payers
Who it's for — Why they bought, in their words
→
fits if: nobody owns distribution
becomes medium once payers' reasons confirm it
How buying happens
→
not if: sells to a buying committee
self-serve product, one decider
What you can deliver now
→
not if: wants strangers messaged automatically
a product rule: it never sends the first message
What they must already have
→
not if: no product yet
needs a URL with something to buy