Learning: Let the interviews rewrite your hypotheses
Synthesize everything your customer interviews have said into concrete updates — hypotheses confirmed, overturned, refined, or newly born — proposed one at a time, with you deciding each. Adapted from Chapter 10 of Hidden Multipliers: A multiplier of one's own.
- Input
- Your GOALS.md, HYPOTHESES.md, and QUESTIONS.md, plus a directory of per-interview debrief files.
- Output
- 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.
What this is about
The interviews were never the point; the learning is. This skill reads your goals, hypotheses, and questions alongside the accumulated debriefs of every conversation you've had, and does what the book says to do with them: notice which hypotheses reality has confirmed, which it has overturned, which need their numbers tuned, and which brand-new beliefs the surprises suggest — then proposes those changes to you one at a time, never in a batch, updating the working files only as you agree.
It holds the book's evidentiary bar: one contradictory voice doesn't change a hypothesis — nothing is universal — but a clear pattern across conversations does, and so does a single genuine revelation. Things heard only once get parked as tentative observations, on the record and watched, rather than acted on. New hypotheses get new interview questions. And when the debriefs start repeating themselves — same stories, same numbers, same language — it tells you the learning has stopped, which means the interviewing should too.
Example invocation
You can invoke the skill like this:
/asb-interview-learning I've done nine interviews since we wrote the
hypotheses — here's the interviews/ directory. What should change?
The skill sweeps your debriefs against your hypotheses and walks you through proposed changes one at a time — each citing the conversations that drive it — updating your files as you agree. The session ends with a concrete verdict: keep interviewing, stop and act, or re-aim.
One practical note: say where you're working. Name a directory (or point at your existing files) when you invoke, and the skill keeps the method's files together there — if you don't, it asks before creating anything.
From the source
Two sources form the foundation of this skill:
- As in Chapter 10 of Hidden Multipliers: A multiplier of one's own (sections "Learning: Iterate the Hypotheses" and "Stop When It's Boring") — hiddenmultipliers.com — the update rules this skill enforces (one voice isn't a pattern; a pattern or a revelation is), the dig-deeper posture, and the stop signal.
- The Iterative-Hypothesis customer development method — the article version: expect contradictions, "no pattern" is still learning, emergent segmentation, market-guru mode, and the warning that AI must not do the thinking for you — which is why this skill proposes and you decide.
Supporting articles each supply a piece of the mechanism:
- Stubborn Visionaries & Pigheaded Fools — the persist-or-stop decision in depth: you can't know which path you're on, so the job is to find out as fast as possible — plus the tactical questions (rate of progress, still learning?, sunk cost, timebox) behind this skill's stop-or-continue verdict.
- The "Convergent" theory of finding truth in darkness — how to tell validated truth from noise: answers that converge across interviews behave like fundamental law; answers that diverge into twenty lobes of dissonance mean the idea, not the interviews, may be the problem.
- "It's a Balance" isn't always the answer — what to do when interviews send contradictory signals: "it's a balance" is usually a refusal to decide — the real options are a clear choice, a maximize-while-satisficing constraint, or inventing "both" by asking what else would have to be true.