Breaking into product: PM interviews
PM interviews test how you think, out loud, under time pressure. This lesson takes each common question type (product sense, execution and metrics, estimation, strategy, behavioural and AI products), shows a structure and a worked answer for each, and ends with a six-week practice plan you can start today.
By the end you can
- Explain what each question type tests and how interviewers score it
- Answer product sense, metric-drop and estimation questions with a structure you can say out loud
- Tell behavioural stories with STAR from a bank of six to eight prepared stories
- Answer AI product questions with problem fit, evals, cost and guardrails
- Follow a six-week practice plan with mock interviews and a scoring rubric
The everyday picture
In a driving test, the examiner does not only check that you reach the end of the route. She watches whether you check your mirrors, signal before turning and slow down when a cyclist wobbles. A driver who arrives safely but never looked in a mirror can still fail, because the examiner had no way to see that they were being careful.
PM interviews work the same way. The interviewer cannot see inside your head, so they score what you say: whether you clarify the goal, choose a user, rank problems, weigh trade-offs and check your numbers. A good answer reached silently scores lower than a decent answer whose reasoning the interviewer could follow. Thinking out loud is the skill being tested.
What PM interviews test, and how they are scored
Processes differ by company and change often, so always read the company's own guidance. A common pattern: a recruiter call, a call with the hiring manager, then three to five interviews of 45 minutes, each on one type of question. Some companies add a take-home task or a presentation. For roles on internal tools or operations products, expect a case about a workflow, stakeholders or a vendor as well.
| Type | Sample question | What they score |
|---|---|---|
| Product sense (design) | "Improve a food-delivery app for rainy evenings." | User focus, choosing and ranking problems, creative but grounded solutions, a success metric |
| Execution and metrics | "Completed trips in one city fell 10% this week. What do you do?" | Structured diagnosis, checking the data first, choosing metrics and guardrails |
| Estimation | "How many delivery riders does a city of 10 million need at dinner time?" | A clear formula, stated assumptions, simple arithmetic, a sanity check |
| Strategy | "Should a ride-hailing app start grocery delivery?" | Market and customer logic, competitors, right to win, risks, a decision |
| Behavioural and leadership | "Tell me about a time you disagreed with a senior stakeholder." | What you did yourself, judgement, results, what you learned |
| AI products | "Design an AI assistant for store managers' questions." | Problem fit, the cost of being wrong, evals, cost and latency, guardrails |
Interviewers usually take notes and rate you against a short list of signals after the interview. Across companies the signals look alike: Do you clarify before answering? Do you pick a user and a problem, with reasons? Do you make trade-offs explicit? Do you use numbers? Is your communication structured and easy to follow? Do you change your mind when given new information? Two habits help in every type: say your structure before you use it ("I'll clarify the goal, pick a user, list their problems, then choose one"), and pause to summarise every few minutes so the interviewer can redirect you.
Quick check
You get the question "Design a product for people who park in a busy city." What should your first minute sound like?
Clarifying the goal and the user is the first signal interviewers look for, and it changes the whole answer. Jumping to a solution skips the reasoning they are trying to score. Ten features with no user or problem shows quantity, not judgement.