The validation playbook
Five stages from picked idea to first paying proof, each with a goal, concrete actions and an explicit kill criterion.
Picking an idea is the cheap part. This playbook is the paid part of the job: five stages that take a dossier you chose and turn it into evidence about YOUR version of it, with your audience, your positioning and your constraints. Every dossier's validation and evidence sections give you the raw material; this page gives you the order of operations.
Two rules before stage one. First, every stage has a kill criterion, and hitting it is a success: you just saved yourself months. Killing an idea after a week of interviews is the system working, and your fit scores will happily rank the next one. Second, do the stages in order. Building a landing page before ten conversations feels like progress and measures nothing.
Stage 1: Restate the bet
Goal: one falsifiable sentence. "People in [audience] pay for [product] because [pain], despite [alternative]."
Do: rewrite the dossier's one-liner in your own words for your specific wedge. Steal the audience section's sharpest segment rather than the broadest one. Write down, before any contact with reality, what evidence would make you QUIT. Sealed predictions keep you honest later.
Kill criterion: you cannot name a specific person who has this pain. Not a persona, a person. If nobody you can find embodies the audience, the audience is a spreadsheet fiction.
Stage 2: Ten conversations
Goal: hear the pain described unprompted, in the audience's words, with what it costs them today.
Do: find ten people from the audience section's watering holes and talk about their workflow, never about your idea. Ask what they tried last time the pain hit, what it cost, who signed it off. The dossier's competition section tells you which alternatives to listen for; if nobody mentions the alternatives at all, the pain may not be felt.
Kill criterion: fewer than three of ten describe the problem without you naming it, or nobody can recall spending money or hours on a workaround. Unfelt pain does not convert, no matter the market size.
Stage 3: The offer test
Goal: a stranger takes a costly action for a product that does not exist.
Do: one page, one promise, one price, one call to action, in the exact words your interviews used. Drive a small, honest amount of the channel the dossier's roadmap suggests first. "Costly action" means an email plus a follow-up answer, a booked call, or a deposit; a like is not evidence. Track by hand, in a spreadsheet - instrumentation is procrastination at this size.
Kill criterion: conversion so far below the revenue section's assumptions that the model breaks even at zero salary. Check the dossier's metrics for what the economics need before you decide what "below" means.
Stage 4: The concierge round
Goal: deliver the outcome manually to three to five real users and watch where the value actually is.
Do: no code beyond duct tape. You are the product: do the research by hand, send the report by mail, run the process in a spreadsheet. Charge something, even a token amount, because free users validate nothing but their own politeness. Write down every step you performed; that list is your build spec, ranked by what users noticed.
Kill criterion: users take the manual result and do not come back, or the manual delivery takes so long that the unit economics can never work even automated. Retention of five concierge users predicts retention of five thousand better than any survey.
Stage 5: First paying proof
Goal: money from someone with no reason to be polite to you.
Do: productize only the steps from stage 4 that users actually valued. Price at the number that made stage-3 prospects hesitate but convert. Ship to the concierge group first, then to the stage-3 waitlist. From here, the dossier's roadmap section is your sequencing document and the score updates on the changelog tell you if the market moved under you.
Kill criterion: the only buyers are friends, or churn after month one exceeds what the revenue model survives. Real strangers, real renewal - that is the bar for calling an idea validated.
Using this with your library
Every dossier ships a validation section with idea-specific experiments and an evidence section with the sources behind the claims - this playbook is the frame, those sections are the ammunition. Deep analyses can pressure-test a shortlist against your own constraints before you commit a single interview hour, and saved ideas alert you when a score moves while you are mid-playbook.