Name the aha

Name the Aha

Find the observable moment after which a user's behavior measurably changes — or conclude, honestly, that no such moment exists yet. Write the verdict to GROWTH.md; every downstream growth skill gates on it.

Gate

None — this is the circle's entry point. If the repo has no GROWTH.md, create one from the bundled template (references/GROWTH-template.md, spec v0.1) before proceeding. The template initializes every gate as closed — that default is load-bearing; downstream skills treat missing gates as closed too.

Inputs

• The repo, and whatever exists of: product copy, onboarding flow, usage data, support tickets, reviews, user interviews.
• If the product can be run, run it — experience the first session yourself before reading anyone's description of it.

Steps

1. Use the product cold. Fresh account, no shortcuts. Note the first moment you felt the thing work — and how many seconds/steps it took. If you can't reach value as a motivated evaluator, note where you stopped.

2. Hunt the behavior change. An aha is not a feeling; it's an event after which usage measurably changes. From any available data, look for the action that separates users who stayed from users who died. No data? Interview transcripts and reviews often name it verbatim ("once I saw X…").

3. Draft the statement. One sentence: user + moment + value. ("A commenter hears their first voice reply" — not "users love seamless audio collaboration.") If it can't fit one sentence, it isn't found yet.

4. Measure the clock. Time-to-aha from signup, measured if possible, estimated-and-tagged if not. Days is unrealistic; hours unrealistic; minutes necessary but insufficient. Set the reduction target one order of magnitude down.

5. Deliver the verdict. Write to GROWTH.md ## Aha (statement, evidence, time-to-aha, target) and set gate: aha = open. Every claim carries an evidence link or the tag hypothesis — an untagged assertion violates the spec.

6. Or refuse. If no behavior-changing moment exists, write NONE FOUND — growth work blocked, keep the gate closed, and output the shortest path to having value instead. This is a success of the method, not a failure of the session.

Conventions

• The aha is the user's moment, not the team's ("first successful deploy", not "first paid conversion").
• If the product's users are agents, the aha is time-to-first-successful-call and the evaluation in step 1 is done through public docs only.
• Never soften the refusal into "the aha is probably X" to unlock downstream skills. A hypothesis-tagged aha is allowed; an invented one is not.

Edge Cases

• Pre-launch product, zero users. Steps 1 and 3 still run; the statement ships tagged hypothesis with a named validation plan. The gate may open on a hypothesis — the tag travels with everything built on it.
• Multiple candidate ahas. Pick the earliest one that predicts retention; log the runners-up in the lore ledger as untested beliefs.
• The team insists the aha is obvious. Write their version down, then do step 2 anyway. Confirmed team intuition costs one query; wrong intuition unexamined costs the whole circle.