Product sense interviews

Product sense interviews test whether you can find a user, name a real pain, propose a solution that fits a company's goals and defend trade-offs under time pressure. Meta calls it product sense; Google often says product design; startups may just say "how would you improve X?" The skill is discovery, prioritization and metrics compressed into ~30–40 minutes of structured talking.

Engineers meet these rounds for PM roles, product-engineer tracks, 0→1 startup interviews and some tech-lead loops. You do not need to be a PM to prepare; you need a repeatable framework and judgment practice. The people who do well sound concrete: named users, named metrics, named cuts — not buzzwords about disruption.

Question types you will see: design ("Design X for Y"), improve ("How would you improve Instagram Stories?"), metrics ("DAU dropped 10%"), occasional estimation and strategy-lite. Prefer company-native strengths (Meta's graph, Stripe's trust) and name v1 non-goals.

Concepts in this section

Read them in order. Each one builds on the one before.

Step	Concept	What it answers
1	Product sense framework	What sequence do I run, and which question type is this?
2	Diagnosing a DAU drop	Dashboard bug, bad release, or a real retention hole?
3	Product sense practice	How do I show judgment, and what do two weeks of drills look like?

Most loops mix a design prompt with a metrics follow-up. If time runs out, skip a third idea before you skip the metric. Interviewers reward decisions: pick a segment by minute 8, force three different solution shapes and choose with a because against your north star.

Deconstruct apps you use weekly — segment, job, metric, one improvement — in 15-minute drills. Record one mock; most people hear they never stated a star or never picked. The same structure helps design reviews and PRDs: clarify outcome, name the user, pick a pain, propose options, choose, define metrics. Prep one real diagnosis story from work (an outage or funnel drop) — authenticity beats invented consumer examples.

Common mistakes

Boiling the ocean, feature salad, no decision. Pick a segment by minute 8; force three distinct solutions; choose.
Skipping verify on metrics questions. Instrumentation and scope before redesign.
Ignoring the company. Use that company's graph, distribution or trust.
Under-waypointing. Say the next step so the interviewer can redirect.

Key takeaways

Know the question types; end design answers in metrics and trade-offs.
Run clarify → goal → segment → pain → solutions → MVP → metrics; waypoint as you go.
For metric drops: verify, scope, split acquisition vs retention, then act today.
Timed mocks and weekly app critiques beat cramming frameworks alone.