PM Interview Question Bank

How would you improve Google Maps?

Hint: CIRCLES — pick one persona (commuter vs traveler) before jumping to solutions.

How would you improve Instagram Reels?

Hint: Optimize for creator novelty, not just viewer time-in-app.

Design a product for elderly users to stay in touch with grandkids.

Hint: Start from the constraints (motor, vision, tech comfort), not the feature list.

How would you improve YouTube for kids?

Hint: Two personas: kids + parents. Trust > engagement.

Design a subscription product for Airbnb.

Hint: Two-sided marketplace: who benefits — host, guest, or both?

How would you redesign the Uber driver app for a new driver in their first week?

Hint: Onboarding is a distinct product from daily use — treat them separately.

How would you improve Netflix's discovery experience?

Hint: Metric matters: "watch minutes" vs "completed titles" vs "return rate."

Design a fitness product for people who hate exercising.

Hint: Reframe: it's not a fitness product, it's a habit-formation product.

How would you improve LinkedIn for job seekers who aren't actively looking?

Hint: Passive candidates need different signals: role fit, trajectory, timing.

Design a product for freelancers to manage taxes.

Hint: Trust and compliance matter more than UX cleverness.

DAU dropped 15% last week. What do you do?

Hint: Segment first (platform / geo / cohort). Distinguish real drop from measurement bug.

You just launched feature X. Metric A went up, metric B went down. Now what?

Hint: Frame as a tradeoff decision, not a bug. Which metric matters more — and why?

How would you diagnose a spike in checkout abandonment?

Hint: Session replay + funnel + external causes (browser update, payment provider).

Your engineering team says a feature will take 3 quarters. Product wants it in 6 weeks. What do you do?

Hint: Break down the WHY behind 3Q. Isolate what a v0 would need.

A key metric shows no movement for 8 weeks. What actions do you take?

Hint: Zero movement usually means: wrong metric, or right metric but no experiment velocity.

Your A/B test result is inconclusive after 4 weeks. Do you ship it?

Hint: Power analysis. What did the pre-registration say? Ship, kill, or run longer are all valid.

How do you prioritize between fixing bugs vs shipping new features?

Hint: Frame in terms of downstream impact on your north-star metric, not moral obligation.

A key stakeholder disagrees publicly with your roadmap. How do you respond?

Hint: Escalate the disagreement, not the personality. Data-driven realignment.

How many pizzas are eaten in NYC in a day?

Hint: Population × % who eat pizza × frequency × pizzas per meal.

How many YouTube videos are watched globally per day?

Hint: Users × sessions per day × videos per session.

How much revenue does the App Store make in a year?

Hint: iOS devices × spend per user × Apple's take rate.

How many Uber rides happen in San Francisco per day?

Hint: SF population × % who Uber × rides per user per week ÷ 7.

How many active Google searches happen per second worldwide?

Hint: Global users × searches per user per day ÷ seconds per day.

How much storage does WhatsApp need for one year of message backups?

Hint: Users × messages per day × avg message size × 365. Sanity-check against S3 pricing.

Should Google enter the cloud gaming market?

Hint: Frame as a strategic bet with a kill signal. What would need to be true?

Should Spotify build its own podcast creation tools?

Hint: Vertical integration playbook: platform vs tool marketplace.

Should Netflix launch an ad-supported tier?

Hint: This has already happened — take a POV on whether it was the right call.

Should Airbnb build a business travel product?

Hint: Two-sided challenge: acquisition on the corporate side, quality on the host side.

How would you think about Stripe entering the consumer banking market?

Hint: Adjacent-market entry: leverage vs risk. Regulatory complexity.

Should Slack integrate deeper with Salesforce products, or stay neutral?

Hint: Post-acquisition strategy: differentiation vs ecosystem play.

Tell me about a time you disagreed with an engineer on a technical decision.

Hint: STAR. Use "I" for the decision, "we" for the outcome.

Tell me about a project that failed. What did you learn?

Hint: Don't hide failure — own the specific decision that caused it.

Tell me about a time you had to say no to a senior stakeholder.

Hint: The interviewer is looking for judgment under authority pressure.

Tell me about a time you shipped something and the metric didn't move. What did you do?

Hint: Owning miss > blaming data. Show the diagnostic loop you ran.

Tell me about the hardest tradeoff you've made as a PM.

Hint: A real tradeoff hurts something you cared about. Show what you gave up and why.

Tell me about a time you influenced a team you didn't manage.

Hint: Concrete tactic: shared metric, weekly sync, 1:1 alignment.

Tell me about a time you launched something faster than expected.

Hint: What did you cut? What did you learn about scope from that experience?

Tell me about a time you had to fire (or push out) a low-performing team member.

Hint: Not for the faint of heart. Interviewer wants to see clarity + care.

Explain how you'd instrument events for a new mobile onboarding funnel.

Hint: Event schema: name, properties, timing, ordering. Watch out for duplicates.

Walk me through how you'd design an experimentation platform for a startup.

Hint: Assignment, exposure logging, guardrails, statistical stopping rules.

What's the difference between latency p50 and p99 — and why does it matter to a PM?

Hint: Tail latency is what users blame. Averages lie.

How would you evaluate an LLM-powered feature before launching to production?

Hint: Eval design: labeled dataset, human raters, offline metrics, then live shadow eval.

When would you choose fine-tuning over prompt engineering for an AI feature?

Hint: Volume, quality bar, latency, and cost — pick your battle.

Explain SQL joins to a non-technical stakeholder.

Hint: Use a spreadsheet analogy. If you can't explain it in 30 seconds, you don't know it.