Product manager product sense interview questions

As asked

Pick any product Google ships. Tell me how you would improve it. We're going to spend most of the conversation here.

Sample answer outline

Pick a clear user segment (e.g. cyclists in dense cities, not 'all users'). State what the current product does well and what it falls short on for that segment. Identify the underlying user need (e.g. cyclists don't want the shortest route, they want the safest route on a given day given traffic and weather). Propose 2-3 concrete improvements and explicitly trade them off. Pick one. State how you'd measure success and what could go wrong. The strong answer is opinionated, structured, and grounded in user empathy - not a feature list.

Expect these follow-ups

• Why this user segment over others?
• What metric would you NOT use?
• What would make you kill this feature six months in?

As asked

Pick an app you use often and tell me one thing you would improve about it. Walk me through who it helps and why it matters.

Sample answer outline

The early-career version rewards structure over polish. Pick one product and one clear user group, describe a specific frustration that group has, and propose one focused improvement rather than a list. Then say how you would know it worked, even a simple metric like more people completing the task. The interviewer wants to see that you start from a real user need, can be specific instead of vague, and can connect a change to an outcome. You do not need to weigh several options against each other yet; one well-reasoned idea grounded in a user problem is enough.

Expect these follow-ups

• Who exactly benefits from this change?
• How would you know if the improvement worked?
• Why this improvement rather than something else?

As asked

A strategic customer says, 'We need a dashboard for operations', but cannot describe the workflow clearly. How do you get from that request to a useful shipped solution?

Sample answer outline

A strong forward-deployed engineer starts by observing the actual work, not collecting a feature wishlist. They identify the users, decisions, source systems, failure modes, and business metric the workflow should improve. The first version should be narrow enough to ship quickly, usually a thin vertical slice with real data and a clear owner on the customer side. The candidate should discuss expectation management, security review, data access, and how the learning feeds back into the core product. A common mistake is becoming a bespoke consultant and building a one-off surface that cannot be maintained.

Expect these follow-ups

• How do you tell whether this should become core product?
• What do you do when the executive buyer and daily users disagree?
• How do you avoid committing to a custom roadmap in the room?

As asked

TikTok is mostly short-form. Should they introduce a long-form video format (10-30 minutes)? Walk me through your reasoning.

Sample answer outline

Clarify the goal: are we trying to grow users, retention, time spent, or revenue (advertiser-friendly long-form has higher CPMs)? Each leads to a different decision. Identify the core risk: long-form may cannibalise the For You Page loop that drives 90% of engagement. Identify the upside: capturing the YouTube use case for creators who already have an audience. Look at evidence from Instagram (Reels vs IGTV) and YouTube (Shorts). Propose a phased rollout - opt-in for top creators, measure cannibalisation, decide. The strong answer is wary of cannibalisation and structured about how to learn.

Expect these follow-ups

• What signal would tell you to roll back?
• How does this change the recommendation system?
• If you had to choose between long-form and live, which would you ship?

As asked

Tell me how you would improve Google Maps, then defend your proposal as a senior PM would: size the opportunity, weigh it against the team's other bets, name the strategic risk, and tell me what you would cut to fund it. I will push on every assumption.

Sample answer outline

The senior version demands prioritisation and strategic judgement, not just a good idea. Open with a sharp user segment and an opinionated improvement, then immediately do the harder work: estimate the size of the opportunity with explicit assumptions, place it against the realistic alternative uses of the team, and state the opportunity cost of choosing it. Name the strategic risk, such as cannibalising an existing surface, an ecosystem dependency, or a defensibility question. Be concrete about what you would not do in order to fund this. The strong answer treats product sense as resource allocation under constraints, holds up under challenge by distinguishing assumptions from facts, and changes its own estimate when a premise is shown to be wrong rather than digging in.

Expect these follow-ups

• What did you assume to size this, and which assumption is weakest?
• What would you stop doing to fund this, and why is that acceptable?
• What is the strongest argument against your own proposal?

As asked

Pick one flow in Google Maps, such as planning a journey or finding a place to eat, and tell me how you would redesign it. As a product designer, focus on the user's experience, the interaction, and how you would validate the new design.

Sample answer outline

Approach it as a designer: start from the user's goal in a specific flow and the friction in the current experience, grounded in how people actually behave rather than a feature wishlist. Walk through the redesigned interaction, the key screens or states, and the decisions that improve clarity, such as reducing steps, surfacing the right information at the right moment, and handling edge states like no results or poor connectivity. Crucially, say how you would validate it: a quick prototype, a usability test with a handful of real users, and the behavioural signal you would watch after launch. The design signal is empathy plus craft plus a habit of testing assumptions with users, not deciding from taste alone, and collaborating with engineering on what is feasible.

Expect these follow-ups

• What does the current flow get wrong for the user?
• How would you test the redesign before building it fully?
• What edge states does this flow need to handle well?

As asked

You have a backlog of fifteen credible ideas and capacity for maybe three this quarter. Walk me through how you prioritize, where a scoring framework helps, and where blindly trusting the score would steer you wrong.

Sample answer outline

Start by anchoring on the goal for the quarter, because priority only means anything relative to an objective. A scoring framework such as reach, impact, confidence, and effort is useful to force consistency and surface the cheap high-impact items, and to make the conversation about explicit assumptions rather than who argues hardest. But the honest part is naming where the score misleads: confidence is often a guess dressed as a number, effort is systematically underestimated, and a framework cannot see dependencies, strategic bets that look weak on a spreadsheet, or must-do work like compliance and keeping the lights on. So I use the score as an input and a tiebreaker, then layer judgment: sequence by dependency, protect a slice for foundational work, and sanity-check that the top of the list actually advances the goal. The signal is using the framework to structure the debate while staying clear-eyed that it is a tool, not the decision.

Expect these follow-ups

• Which input in a scoring framework is the least trustworthy, and why?
• How do you handle a strategically important item that scores poorly?
• Where do dependencies fit, since the score ignores them?

As asked

Analytics show a big chunk of new users abandon during onboarding before they ever reach the core product. How do you investigate where and why they drop, and how do you decide what to change first?

Sample answer outline

Begin by making the funnel concrete: instrument each onboarding step and find where the largest drops happen, because the fix for an early drop, often friction or unclear value, is different from a late one, often a hard ask like permissions or payment. Pair the quantitative funnel with qualitative insight, watching session recordings or talking to a few users who bailed, since the numbers tell you where but rarely why. Form a hypothesis for the biggest leak: too many steps before the user sees value, an account wall too early, a confusing screen, a slow or broken step on mobile. Then prioritise by impact and effort, usually attacking the largest drop with the cheapest credible fix first, and validate with an experiment rather than assuming the change helped. Frame success as users reaching the activation moment, the point where they first get real value, not just finishing the flow. The signal is moving from where to why to a tested change, anchored on getting users to value sooner.

Expect these follow-ups

• Why does the location of the drop change what you should fix?
• How do you combine the funnel data with qualitative insight?
• What does activation mean here, and why measure it over flow completion?

As asked

A new project management app shows an empty dashboard after signup. Write the empty state strategy, not just the headline.

Sample answer outline

The empty state should orient the user, explain why the space is empty, and provide one primary next action. The copy should reflect the user's stage: a first-time user needs setup guidance, while an established user with no tasks may need reassurance or a filter reset. Avoid generic cheerleading and avoid presenting three equal calls to action. Pair the message with an affordance such as creating the first project, importing tasks, or using a template. Senior candidates connect the copy to activation metrics and onboarding research rather than treating it as decoration.

Expect these follow-ups

• What would you change for an enterprise admin setting up a team?
• How do you decide whether to use a template gallery or one primary button?
• What metric would tell you the empty state is working?

As asked

Developers keep raising the same complaint in Discord, GitHub issues, and conference hallway conversations. How do you turn that signal into a product change?

Sample answer outline

Good DevRel work converts anecdotes into usable product signal. Cluster the complaints, identify affected personas, collect reproduction steps or examples, and measure scale using support tickets, docs search, issue volume, or funnel drop-off. The candidate should produce a concise product brief with evidence, impact, and possible fixes, then work with PM and engineering on priority. They should also close the loop publicly when something changes. A weak answer treats community volume as automatically representative without checking severity or business impact.

Expect these follow-ups

• How do you handle one loud user who may not represent the market?
• What makes feedback actionable for engineering?
• How do you communicate that a requested change will not be built?

As asked

Adobe is integrating generative AI features like Firefly into Photoshop and Illustrator. As an engineer on that team, what are the biggest technical and product tradeoffs you need to navigate? Consider latency, cost, copyright, trust, and the experience of professional users.

Sample answer outline

Strong answers cover at least four dimensions: latency (inference is slow; users expect immediate feedback; solve with progressive previews, speculative decoding, or cached style seeds), cost (GPU inference at scale is expensive; token budgets, model distillation, and client-side inference for simpler tasks matter), copyright and provenance (Adobe's Content Credentials and C2PA standard are relevant here; professional users care about liability), and trust (professionals need to understand what the AI changed so they can correct it; non-destructive AI layers and prompt history are more trustworthy than magic black boxes).

Expect these follow-ups

• A professional photographer is worried that AI-generated content will devalue their work. How does Adobe's approach address that?
• Your AI feature is 3x slower on older hardware. How do you decide what minimum spec to support?

As asked

What is Amazon's Working Backwards process, and how would you apply it if your team were scoping a new feature for Alexa's home automation integration? Walk me through what a press release and FAQ would look like for that feature.

Sample answer outline

Working Backwards starts with a press release written from the customer's perspective describing the finished product, then a FAQ addressing external questions (what is it, who is it for, why does it matter) and internal questions (what are the hardest technical problems, how do we measure success). For the Alexa home automation example, the press release should open with the customer benefit, not the technology. The FAQ should cover the customer's current pain (fragmented apps per device, unreliable voice control), the new experience, and how success is measured (time to complete a home-control task, retention of the feature after 30 days).

Expect these follow-ups

• What is the most common mistake teams make when writing a PR/FAQ?
• How does a PR/FAQ differ from a traditional product requirements document?

As asked

TikTok wants to improve creator monetization to retain top creators who are being poached by competitors. You have 6 months and a team of 8 engineers. What do you build and how do you decide?

Sample answer outline

A strong answer starts by segmenting creators (mega, mid-tier, micro) and identifying where the retention problem is worst. The candidate then articulates the current monetization landscape (TikTok Creator Fund, brand deals, live gifts) and its known gaps (creator fund pays poorly per view). They propose one or two concrete bets with reasoning: for example, a subscription model or a brand marketplace that matches mid-tier creators with advertisers. They define a success metric (creator 6-month retention rate by tier) and acknowledge the platform tension between creator monetization and advertiser relationships.

Expect these follow-ups

• How do you prioritize between retaining your top 1% creators versus building tools for the long tail?
• What is the risk of your proposed solution and how would you mitigate it?

As asked

You are a PM for Google Search. How would you define and measure the quality of search results? What metrics would you look at, how would you set up a measurement framework, and how would you distinguish between short-term and long-term quality signals?

Sample answer outline

A strong answer separates relevance (did the top result answer the query?), freshness (is the information current?), diversity (does the result set cover multiple intents?), and user satisfaction proxies like long-click rate, reformulation rate, and session abandonment. The candidate should mention the distinction between direct metrics (NDCG from human raters), behavioral proxies (click data), and long-term signals (return visits, query volume growth). Google uses human quality raters with defined guidelines, so an aware candidate mentions this.

Expect these follow-ups

• How do you handle queries where there is no clear 'right answer', like opinion-based searches?
• How would you detect if a short-term metric improvement is actually harming long-term user trust?

As asked

LinkedIn is testing a new format for sponsored posts in the feed. You want to measure whether the new format increases both click-through rate and member satisfaction (proxied by post-click dwell time), without cannibalizing organic engagement. How would you design this experiment?

Sample answer outline

A strong answer covers unit of randomization (member-level to avoid contamination), sample size calculation based on expected effect size, holdout groups, and the guardrail metrics (organic engagement rate, session length, complaint rate). The candidate should also discuss novelty bias, the tradeoff between running the test long enough for novelty to wear off versus business urgency, and how to handle multiple testing corrections across the two primary metrics.

Expect these follow-ups

• How do you decide when the test has run long enough to make a call?
• What would you do if CTR improved but dwell time dropped?

As asked

The Instagram feed ranking team wants to improve time-spent as a metric. You are running an experiment with a new ranking signal. How do you design the experiment, choose metrics, and decide whether to ship?

Sample answer outline

A strong answer frames the guardrail metrics (likes, comments, shares, follows) alongside the primary metric (time-spent), explains why time-spent alone can be gamed by addictive bad content, and designs an A/B test with holdout sizing based on minimum detectable effect. The candidate should mention how to handle network effects in feed experiments (users on the same social graph can contaminate control and treatment), and what Novelty effect means for ramp duration.

Expect these follow-ups

• Your experiment shows time-spent up 3% but shares down 5%. Do you ship?
• How do you measure long-term user satisfaction when your experiment only runs for two weeks?

As asked

Data shows that 30% of Netflix subscribers open the app, scroll for a few minutes, cannot find something to watch, and close the app without playing anything. What product changes would you explore to fix this, how would you prioritize them, and how would you measure success?

Sample answer outline

A strong answer first diagnoses the root cause (poor personalization signal, overwhelming choice, low-quality thumbnails, wrong surface for mood-based browsing) before jumping to solutions. Solutions might include a mood-based filter ('I want something funny and short'), a continue-watching nudge, a social proof layer showing what friends watched, or a direct 'Play Something' button for decision-fatigued users. Measurement should cover play rate from browse sessions, not just clicks on a feature, and the candidate should distinguish between engagement metrics and the long-term retention metric Netflix actually cares about.

Expect these follow-ups

• How do you balance showing users content they are likely to watch versus content they might be surprised by and love?
• What would a bad metric for this problem look like, and why might it be tempting to use it?

As asked

ChatGPT has a memory feature that lets it remember facts about you across conversations. If you were the engineer responsible for this feature, what would you measure to know if it is working well, and what would you change to make it better?

Sample answer outline

A strong answer starts with metrics: recall accuracy when relevant memories exist, precision when memories are injected, user satisfaction signals (do users feel understood vs surveilled), and memory staleness. Then improvements: surfacing what the model has remembered and letting users edit it, smarter deduplication, handling contradictory memories, privacy controls and geographic compliance, and selective injection (not always injecting all memories into every prompt).

Expect these follow-ups

• How would you handle the case where a stored memory becomes outdated or wrong?
• How do you balance personalization with the cost of injecting memory into every prompt?

As asked

Your squad wants to ship a new personalization feature that changes how the Home tab orders content for users. Walk me through how you would design an A/B test, what metric you would use as the primary success metric, what guardrail metrics you would watch, and how long you would run the experiment before making a call.

Sample answer outline

A strong answer starts by picking a north-star metric tied to Spotify's business, such as weekly active listening minutes or track completion rate, and explains why that is better than CTR alone. It describes how to randomly split users, what sample size is needed to detect a meaningful lift (power calculation), and what guardrail metrics (like subscription cancellations or support tickets) would block a rollout even if the primary metric is green. It addresses novelty effects and how long to run to avoid them.

Expect these follow-ups

• How do you handle the fact that users in the control and treatment groups talk to each other and influence listening behavior?
• What would you do if the experiment shows a positive lift for premium users but a negative one for free users?

As asked

You run an A/B test on a new tweet ranking feature. The treatment group shows a 4% lift in daily engagement (likes, replies, retweets) but also a 12% increase in content reported as harmful or abusive. The test is statistically significant on both metrics. What do you do?

Sample answer outline

A strong answer refuses to ship based on engagement lift alone, frames the decision around long-term user trust and platform health, and proposes decomposing the 12% harm increase to understand what kind of content is driving it (is it the ranking surfacing existing harmful content or generating new user behavior?). The candidate should recommend not shipping until a harm-neutral variant is found, escalate to a safety review if needed, and recognize that pure engagement metrics incentivize the system to surface outrage. X has historically struggled with this trade-off, so the candidate should show nuanced judgment rather than a simplistic 'do not ship' answer.

Expect these follow-ups

• Your manager says the revenue impact of the engagement lift is too large to ignore and pushes you to ship. How do you respond?
• How would you redesign the experiment to measure user value rather than just engagement?

As asked

Only 40% of users who start the Airbnb host onboarding flow complete it and publish a listing. You are given no additional budget and 3 engineers for 6 weeks. What do you do?

Sample answer outline

Strong answers start by diagnosing where in the funnel people drop off (step-level funnel analysis), then segment by device, experience level, and listing type to find the worst-performing cohort. They propose a focused experiment on the highest-drop step rather than redesigning everything, measure both completion rate and listing quality (hosts who onboard faster may list lower-quality properties), and have a clear shipping criteria before starting.

Expect these follow-ups

• You run an experiment showing completion rate goes up 12% but quality score of new listings drops 8%. Do you ship?
• How would you use qualitative research (user interviews, session recordings) to complement the funnel data?

As asked

Anthropic makes money through both the API and the Claude.ai consumer subscription. These two businesses have different growth levers, different customers, and sometimes different safety requirements. How would you think about prioritizing investment between them if you were leading product strategy?

Sample answer outline

A strong answer frames the API as the higher-margin, more defensible business in the long term because developers build on it and switching costs are high, but the consumer product provides direct feedback on real-world use, brand recognition, and revenue diversification. It should address the tension: enterprise API customers want stability while consumer requires rapid feature iteration. The safety requirements differ: consumer surfaces are open to anyone including vulnerable populations, while API customers have agreed to usage policies. A thoughtful answer notes that Anthropic's mission argument favors making safe AI accessible broadly, which the consumer product serves.

Expect these follow-ups

• How would you decide whether to ship a feature on Claude.ai before it is available on the API?
• If the consumer product is losing money and the API is profitable, how do you make the resource allocation argument?

As asked

During bad weather or a major sporting event, Dasher supply drops and order volume spikes. Walk me through how you would design a dynamic pay boost system that incentivizes more Dashers to come online, and how you would communicate it to Dashers in the app.

Sample answer outline

Strong answers model the supply-demand imbalance per zone in real time, set boost amounts using a curve fitted to historical data on Dasher responsiveness to pay increases (price elasticity), and avoid sudden large swings that confuse Dashers. They address the communication design: push notification with expected boost amount, app UI showing the zone and time window, and transparency about when the boost ends. They should also mention that a too-large boost attracts Dashers from neighboring zones, disrupting those zones.

Expect these follow-ups

• How would you A/B test whether a $2 boost vs a $3 boost produces better supply-demand balance?
• What guardrails would you put on the maximum boost to protect DoorDash's unit economics?

As asked

You're the PM for GitHub Actions. The team has capacity for one large feature next quarter. Candidates include: better test result visualization in PRs, a re-runnable failed-jobs-only option, support for GPU runners, and workflow debugging tools. How do you decide what to build?

Sample answer outline

Describe a framework: quantify which feature unblocks the most users (breadth), how painful the gap is today (depth), and how it ties to GitHub's competitive position (strategic fit). Failed-jobs-only reruns has broad applicability and is a high-frequency frustration; GPU runners serve a smaller but high-value ML segment. Strong candidates propose how they'd validate their prioritization (user research, usage data, competitor analysis) before committing.

Expect these follow-ups

• How would you measure the success of whichever feature you ship three months after launch?
• A large enterprise customer is pressuring you to build a feature they need but that only benefits 0.1% of users. How do you handle that?

As asked

GitLab has been shipping AI features under the GitLab Duo brand: Code Suggestions, Chat, vulnerability explanation, and others. If you were a product engineer on that team, which feature would you prioritize improving and what specific change would you make? Walk me through your reasoning.

Sample answer outline

Strong answers pick a specific feature and a specific improvement grounded in user pain, such as Code Suggestions latency being too high for adoption in fast-paced editing, or Chat not having context about the current file. They should propose a concrete change (streaming completions, repo-level context indexing) and describe how they would measure success.

Expect these follow-ups

• How would you decide between improving suggestion quality vs. improving latency?
• How would you measure whether the improvement actually changed developer behavior?