45-minute mock plan
- 5 minutes: choose one product case and write the business decision at the top of the page.
- 10 minutes: answer out loud without notes. Record only the opener, metric cuts, recommendation, and caveat.
- 10 minutes: rewrite the opener so it starts with the decision and the metric grain.
- 10 minutes: answer the follow-up: what would change your recommendation?
- 10 minutes: repeat the answer once, faster, with fewer charts and a clearer operating call.
Mock prompt
A marketplace product has strong signup growth from a new channel, but week-two retained users are flat. The PM asks whether to double spend on that channel. What metric view do you build first, and what do you recommend?
Answer shape
- Start with the decision: keep spend flat, double spend, narrow spend to a segment, or pause until quality is proven.
- Define the grain before diagnosing: one row per user, signup, seller, buyer, session, order, or week-two cohort.
- Cut the channel by intent, first successful action, acquisition cost, role, geography, and week-one activation.
- Separate growth from quality. More signups do not matter if the channel brings low-intent users who never reach the marketplace action that predicts retention.
- End with one recommendation, one risk, and one metric that would change your mind.
What to avoid
- Do not recommend more spend because signups are up.
- Do not skip retention definition, eligibility, or cohort timing.
- Do not hide behind "more data needed" without naming the next cut.
- Do not treat SQL as separate from product judgment. The query should prove or disprove the business call.
Where the packet helps
The Product Analytics packet adds worked answers, follow-up challenges, IC5 vs Staff+ comparisons, metric architecture templates, and an AI-native practice setup you can run against your own sanitized projects.