A Claude skill that critiques product ideas, UIs, and docs using evidence from 303 long-form interviews with SaaS operators, not generic growth advice.
Most growth advice repeats the same handful of platitudes. This skill is a distilled corpus instead: 303 interviews with operators at Reforge, Superhuman, Segment, HubSpot, Atlassian, Wistia, GoDaddy, ConvertKit, Bubble, Pinterest, and around 290 other companies, turned into 9 reference files covering onboarding, activation, churn, retention, win-back, pricing, acquisition, customer success, plus a file of the sharpest operator voices and a file of the least obvious tactics.
Every claim in it traces to a named person at a named company. No claim is left as a vague "studies show."
Point it at one of three things:
- A product idea, a P2, or a doc — it critiques: what the evidence supports, what it contradicts, what nobody has tested.
- A UI or a flow — it tears it down: walks the surface like a customer, finds where it leaks, and cites the operator evidence for why each leak costs money.
- Something that exists plus a goal — it proposes improvements, ranked by strength of evidence, each one with the condition under which it stops working.
If the corpus doesn't cover a question, it says so instead of filling the gap with generic advice.
- Cite person and company, always. Never "one founder" or "studies show."
- State the condition under which a mechanism inverts. A tactic without a boundary is a liability.
- Keep every number's denominator and population attached.
- Flag dated or dubious claims instead of repeating them as current truth.
- When several operators independently land on the same mechanism, say so — that's the strongest evidence class in the corpus.
Drop SKILL.md and references/ into your Claude skills directory. See SKILL.md for the full routing table and rules of engagement.