How do you prepare for a product manager interview?

The question types

Product sense. Design something for someone, improve an existing product, decide between options. Testing whether you think about users before features.

Analytical / metrics. A number moved, work out why. Or: what would you measure for this feature. Testing whether you can decompose a metric and reason about causes.

Execution. Prioritisation, tradeoffs, what you would cut. Testing judgment under constraint.

Behavioural. Your actual past. Testing whether the answers above describe things you have really done.

Strategy, at senior levels. Should the company enter this market, how should it respond to a competitor.

Structure without recitation

You need enough structure to stay organised out loud and not so much that you sound like you are filling in a form. A workable spine for product sense:

1. Clarify the goal and constraints. Ask one or two real questions, not five stalling ones.
2. Pick a user segment and say why that one.
3. Name their problems, then pick the one worth solving.
4. Propose two or three genuinely different solutions.
5. Choose, with a stated reason.
6. Say how you would measure it, and what would tell you that you were wrong.

Step six is the differentiator. Almost nobody volunteers the conditions under which their idea fails, and it is the most senior thing you can do in an interview.

Your stories matter more than your frameworks

Prepare two or three situations in genuine detail: the context, what you actually decided, what the tradeoff was, what happened, what you would do differently. Interviewers probe, and stories rehearsed at headline level collapse on the second follow-up question.

Include one where you were wrong. Candidates avoid these and the avoidance is obvious; a well-told failure with a real lesson is more convincing than another success.

The most useful preparation is to write the stories out and then delete the parts where you sound blameless.

Know the product properly

Use it. Sign up, complete the core flow, hit a rough edge. Then form an opinion you can defend about one thing you would change and why — including who it would be worse for, since every change has a loser.

Candidates who have clearly not used the product are memorable in the wrong way, and the bar here is low enough that clearing it is genuinely differentiating.

Reason out loud on metrics questions

When asked why a number dropped, resist guessing. Segment it: is it all users or one group, one platform, one geography, one channel. Is it a measurement change or a behaviour change. Is it seasonal.

Working through the decomposition audibly is the answer. Intercom's jobs-to-be-done thinking is useful background here — the strongest candidates keep returning to what the user was trying to accomplish rather than to the feature under discussion.