Product manager interview questions, by type

What is your favorite product, and how would you improve it?

Interviewers sometimes ask what you'd remove from the product. Having a view on that shows you see its trade-offs.

Google: How would you improve Google Maps for tourists?

A likely follow-up: a tourist who can't read the local language. Have a sense of how your idea works for them.

Meta: How would you help new creators grow on Instagram?

What does growing mean to a new creator: reach, income or simply getting replies? Each leads to a different product.

How would you improve Spotify's onboarding for new users?

New users arrive from different places, such as a friend's playlist, a free trial or a new phone. Where they came from shapes what onboarding should do.

How would you improve LinkedIn for people looking for their first job?

Many first-job seekers are students on a school calendar. What they need in September differs from what they need in April, and timing can be part of the design.

Design an alarm clock for heavy sleepers.

Decide early whether this is an app, a device or both. That choice sets what's possible and what a first version costs.

Design a product that helps elderly people stay connected with their families.

Health and hearing change from day to day. Think about how the product behaves on a day the older person isn't feeling well.

Design an expense reporting tool for employees at a 500-person company.

Companies buy these tools, and a pilot team usually tries one first. A good answer says how the design wins that team over before the whole company sees it.

You just launched one-click checkout in a shopping app. How would you measure whether it's a success?

Some users already had a fast checkout through saved cards. The feature may change little for them and a lot for others, which affects how you read the results.

Instagram Reels watch time dropped 10% week over week. How would you investigate?

Interviewers often add facts mid-answer, such as "it's only on Android". Showing how each new fact changes your next step is the point of the exercise.

Meta: How would you define and measure a healthy Facebook Group?

A group that looks quiet can still work well, like a neighborhood group that posts once a week. A good measure doesn't punish it.

Airbnb: What would you choose as Airbnb's north star metric, and why?

Some stays last one night and some last a month. Whether your metric treats them the same way, and whether it should, is worth raising.

Your A/B test raised checkout conversion by 3%, but average order value fell 4%. Do you ship it?

Ask what the company is chasing this quarter. A team focused on first purchases reads this result differently from one focused on margin.

Your team wants to send more push notifications to raise daily active users. What guardrail metrics would you set before launching?

Light users and heavy users react differently to extra messages. A guardrail averaged over everyone can hide damage in one group.

How many rides does Uber do in New York City on a typical weekday?

Small scope choices, like what a typical weekday is and whether airport trips count, move the answer more than the arithmetic does.

How many smartphones are sold in the US each year?

Say your plan in one sentence before the first number, so the interviewer can follow it and correct you early.

Estimate how much new storage YouTube needs every day.

If you know little about video files, say so and reason from something you do know, like the size of a video on your phone.

Estimate the yearly revenue opportunity for a dog-walking app in the US.

Dog walking is mostly a city service. Adjusting for where people live is what makes the estimate believable.

You have three features your stakeholders want and engineering capacity for only one this quarter. How do you decide?

A cheap way to learn about the two you didn't pick, such as a quick test or a few customer calls, makes next quarter's choice easier. Mention it.

Two weeks before launch, engineering tells you the launch will slip six weeks. What do you do?

A follow-up to expect: what if the date is tied to a public event you can't move? Have an answer for that version too.

Your biggest customer says they'll leave unless you build a feature that doesn't fit your roadmap. What do you do?

What you tell the customer this week, before anything is decided, is part of the answer. Holding the relationship while you work it out takes skill.

You've just joined as the PM of an existing team. How would you set its goals for next quarter?

The team already had goals before you arrived. Interviewers notice whether you'd keep what works instead of resetting everything in month one.

Tell me about a time you had to prioritize between requests from several important stakeholders.

Pick a story where every stakeholder was senior and had a fair case. It's harder to tell, and better, than one with an obvious winner.

Should Netflix launch a live sports tier?

Live sports bring a fixed schedule and a different kind of viewer. What Netflix would have to build to run live events well belongs next to what it would pay.

A US ride-sharing company wants to launch in India. How would you decide whether and how to enter?

Expect follow-ups on drivers as much as riders. In a new country, supply is often the harder side.

How would you think about a restaurant delivery app expanding into grocery delivery?

Think about the customer's week: when they'd order groceries and when dinner, and whether they want one app for both.

You lead product at a meditation app. Should you build your own sleep-tracking wearable?

A useful follow-up to prepare: what would you do if a large rival launched a similar device next month?

An AI assistant in your app gives a confidently wrong answer in about 3% of responses. What do you do?

Can users tell when the assistant is wrong? Confident wrong answers do more harm than ones that sound unsure, so start from what the user sees.

Users love your new AI feature, but each request costs you $0.40. How do you decide what to do?

Model prices change quickly. A strong answer says what you'd do now and what would make you revisit it in six months.

Your AI feature takes 8 seconds to respond. Users say it's too slow, but the faster model gives worse answers. How do you decide?

How would you hear that it's too slow: complaints, people leaving mid-answer or a survey? Each tells you something different.

You're building a support bot for a software company's help center. Would you use prompting, retrieval, or fine-tuning?

Define each option in one plain sentence before comparing them. Interviewers check that you know what each one changes, then how you'd decide.

Explain what happens when you type a website address into a browser, the way you'd explain it to a non-technical executive.

This tests whether you can switch levels on demand. Try it at three lengths: one sentence, one minute and five minutes.

At a PM level, how would you design the system behind a "people you may know" feature on a social network?

The awkward cases, like suggesting an ex or a coworker someone avoids, say a lot about your product judgment.

Tell me about yourself.

A PM and a recruiter listen for different things. Have a version ready for each.

Why product management?

Say why now as well as why PM. Interviewers want to know what changed that makes this the right moment.

Tell me about a product you shipped that you're proud of.

Choose a launch you can discuss for twenty minutes, since follow-ups will dig into design, data and people.

Tell me about a time you disagreed with an engineer on your team. How did you resolve it?

Choose a recent story with real stakes, and say how long it took to settle. A disagreement resolved in one meeting can sound too easy.

Tell me about a time you had to get a team to do something when you had no authority over them.

Influence usually builds over several conversations, so how long it took belongs in the story and makes it easier to believe.

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

Interviewers often ask what you'd do differently with only the information you had then. Separating bad luck from a bad call shows judgment.

Amazon: Tell me about a time a metric looked wrong and you dug into the details yourself to find out why.

Dive Deep is one of the Leadership Principles Amazon publishes. Read its short description on Amazon's site and choose a story that matches it closely.