Skip to content
Skills from A Smart Bear

Customer Interviews

The Iterative-Hypothesis customer development method

Six skills, run in sequence, that turn customer conversations into decisions you can defend — what to build, who for, what to charge, and how to say it. Adapted from The Iterative-Hypothesis customer development method.

What this is about

Most customer interviews are wasted — undirected conversations and leading questions that end up confirming what you wish were true instead of uncovering what actually is. This workshop is the antidote: a method Jason Cohen has honed over twenty years and used to validate entire startups (WP Engine among them) and to find the incremental features people will actually pay for. It's the "How to Extract Insights from Customers" procedure from A multiplier of one's own, the final chapter of Hidden Multipliers, and the published Iterative-Hypothesis customer development method.

The insight that makes it work: you can't ask customers what you actually need to know. You can't ask "What should we charge?" or "Would you buy this feature?" — the first gets a shrug, the second gets a polite "yes" that evaporates when the invoice arrives. So instead you decide privately what you need to learn, write down what you already believe, and then design open-ended questions that let customers describe their world while you draw the conclusions. Do that across enough conversations and clear patterns emerge — the real prices, the real objections, the real words.

The six skills below are that method, broken into steps you run in order. Each produces a file the next one consumes, so the work compounds instead of scattering across notebooks and memory.

How to run it

  1. Set the goals. Start with Goal Questions — the unaskable questions (price, positioning, ideal customer, what to build) your interviews must answer. This writes GOALS.md, and every later step traces back to it.
  2. Write down what you believe. Feed those goals to Hypotheses, which records your current best guesses as numbered, falsifiable statements in HYPOTHESES.md. Writing predictions before you test them is what makes you learn fast instead of quietly rationalizing whatever you hear.
  3. Turn hypotheses into questions. Interview Questions converts each hypothesis into an open-ended question that tests it without leading the witness, producing QUESTIONS.md — your interview script.
  4. Interview a customer, then file it. After each conversation, run Interview Debrief to distill the transcript or your notes into one short per-person record, mapped to your questions, with the surprises kept verbatim.
  5. Let the interviews rewrite your beliefs. Once you've stacked up a few debriefs, Learning reads them against your hypotheses and proposes updates one at a time — confirm, overturn, refine, or spawn a brand-new hypothesis (and the question to test it). Then you go back to step 4 and interview more people.

Keep looping steps 4 and 5 until it gets boring. There's no magic number — ten interviews validated some ideas; WP Engine took forty. When you keep hearing the same stories, the same numbers, and the same language, the surprises have stopped, which means the learning has stopped. That's the signal to stop gathering and start deciding.

What you get at the end

When the interviews are done, Interview Report distills everything — goals, hypotheses, questions, and all the debriefs — into a single FINAL-REPORT.md: a brief summary anyone on the team will actually read, backed by numbered, evidence-cited findings and per-area briefs for ideal-customer, positioning, pricing, marketing, and product. That report is the validated set of facts you build the next round of decisions on.

The files, at a glance

The sequence is really a chain of files, each handed to the next step:

GOALS.md → HYPOTHESES.md → QUESTIONS.md → a growing folder of per-interview debriefs → (Learning edits the hypotheses and questions in place, round after round) → FINAL-REPORT.md.

The steps

This workshop runs several skills in sequence, and some rely on others (like asb-rude-qa) — install the whole set before you begin: run npx skills add asmartbear/asb-skills, or ↓ download all skills as one ZIP. More install options →

  1. Goal Questions

    Defines what your interviews must answer.

    Your business — new idea or existing company — and the decisions you want customer interviews to settle. A GOALS.md file: your context in prose, then numbered goal questions (G1, G2, …) your interviews must answer.

    asb-interview-goals

  2. Hypotheses

    Turns your assumptions into testable bets.

    Your GOALS.md goal questions, plus what you currently believe about customers, pain, price, and competition. A HYPOTHESES.md file: numbered, falsifiable hypotheses (H1, H2, …), each mapped to the goal questions it addresses.

    asb-interview-hypotheses

  3. Interview Questions

    Forges unbiased questions from your hypotheses.

    A HYPOTHESES.md file — or a single hypothesis you want to test in an interview. A QUESTIONS.md of numbered questions in interview order, each mapped to the hypotheses it tests (single-hypothesis mode: the final question, in chat).

    asb-interview-questions

  4. Interview Debrief

    Distills one interview into structured evidence.

    Raw material from one customer conversation — a transcript, your scribbled notes, or a voice-memo dump — plus your QUESTIONS.md. A brief per-person debrief file in your interviews directory: one concise answer per question asked, and an addenda of things you heard that fit no question.

    asb-interview-debrief

  5. Learning

    Turns interview patterns into hypothesis updates.

    Your GOALS.md, HYPOTHESES.md, and QUESTIONS.md, plus a directory of per-interview debrief files. Your working files updated one agreed change at a time — plus tentative observations worth watching and a read on whether it's time to stop interviewing.

    asb-interview-learning

  6. Interview Report

    Distills your interviews into usable truth.

    Your GOALS.md, HYPOTHESES.md, and QUESTIONS.md, plus the directory of per-interview debrief files. A FINAL-REPORT.md: a brief top summary humans will actually read, backed by numbered, evidence-cited findings and per-area briefs for ICP, positioning, pricing, marketing & sales, and product.

    asb-interview-report