Product Sense

What this round is

Product Sense is the round where the interviewer hands you a product or product space and asks you to think out loud. The format is usually open-ended: "How would you improve Instagram?" or "Design a product for X." There is rarely a single correct answer. The interviewer wants to see how you frame the problem, who you build for, what trade-offs you weigh, and whether your conclusion is something a thoughtful PM could actually defend.

What they're testing

Not your ideas. Your judgment. Anyone can list features for Instagram in five minutes. The candidates who clear this round are the ones who choose a user, name a real problem that user has, propose a solution that connects to the problem, and explain why this solution beats other plausible options. The interviewer is watching for structure, prioritisation, and the ability to defend a choice without being defensive.

The four things being scored:

1. Framing - can you turn an open question into a specific one?
2. Empathy - do you actually know who you're building for?
3. Prioritisation - can you choose between plausible options and defend the choice?
4. Communication - can the interviewer follow your thinking without effort?

The mental model - 7 questions

Most interview prep teaches Product Sense as a framework - a sequence of steps you recite. The 8-step version is below if you want it. But senior PMs don't recite. They think. And what they're thinking is a small set of questions, asked in whatever order the conversation demands.

Don't memorise a framework. Internalise these seven questions. If you can answer them in any order with rigour, you'll outperform 90% of senior candidates regardless of which framework they ran.

1. What are we actually solving for? Restate the prompt in your own words. Confirm the user, the goal, and the constraints. Strip away the noise.
2. Why does this matter strategically? Connect the problem to the company's bigger picture. What does solving this unlock - for the user, for the business, for the long-term position?
3. Whose pain is sharpest, and what's the non-obvious truth? Name a specific user. Identify the real problem - not the surface complaint. The non-obvious insight is what separates a good answer from a great one.
4. What are the genuinely different ways to solve this? Generate two or three solutions that are structurally different from each other. Not three variations of one idea. Different angles, different bets.
5. What's the right bet, and what am I deliberately not doing? Pick one. Defend the choice. Name what you're explicitly choosing not to do - and why.
6. What does v1 look like, and how will I know it worked? Define the minimum viable version. Name the metric that tells you it's working. Be specific.
7. What could break, and where does this go in 3 years? Name the risks honestly. Name the second-order effects. Show that you're thinking past the launch.

The order doesn't matter. The rigour does.

The formal framework - 8 steps

If you're earlier in your career or want a more structured scaffold, there's a formal 8-step version. It maps onto the 7 questions above, just with more granular sub-steps and acronyms to help you remember them under pressure.

Use whichever is more useful for you. The 7 questions are the destination. The 8 steps are training wheels - useful when you're learning, optional once you're senior.

Sample questions

Product Sense questions usually fall into one of three shapes - improve, design, or evaluate. The framework above works for all three; the framing of the answer shifts slightly depending on which shape you're handed.

Improve an existing product.

• Improve Instagram Stories.
• Improve YouTube recommendations.
• Improve LinkedIn job search.
• Improve Airbnb host reliability.
• Improve Amazon reviews (reduce fake reviews).
• Improve Google Maps navigation for commuters.
• Improve WhatsApp groups.

Design a new feature.

• Design a feature for WhatsApp to reduce noise in groups.
• Design a feature for Spotify to help users discover relevant content.
• Design a feature for LinkedIn to help freshers stand out.
• Design a feature for Google Calendar for families.
• Design a feature for Amazon to reduce cart abandonment.
• Design AI summaries for Spotify podcasts.

Design for a user segment or zero-to-one.

• Design a banking app for senior citizens.
• Design a productivity tool for ADHD users.
• Design Instagram for visually impaired users.
• Design a product to reduce loneliness.
• Design an AI tutor for kids.
• Design a product that helps people sleep better.

How to structure your answer

The structure of your answer is half the signal. A candidate who jumps straight to features loses the round even if the features are good. A candidate who walks through the 7 questions in some order, transparently, wins even if the final idea is unremarkable.

• Talk while you think. The interviewer needs to follow you. Silent thinking signals lack of structure.
• Use the whiteboard or screen. Sketch the user, the problem, the options. Visual structure beats verbal.
• Number your sections. Say "first I'll do X, then Y, then Z" at the start. Stick to it.
• Time yourself loosely. Spend ~5 minutes on Questions 1-3 (framing and user), ~10 on Questions 4-5 (options and choice), ~10 on Questions 6-7 (v1 and risks).
• Recover gracefully. If you realise mid-answer you picked the wrong user, say so. Backtrack visibly. Senior PMs course-correct in real time.

Common traps

• Listing features instead of solving a problem. The fastest way to lose this round.
• Picking "everyone" as your user. A segment that includes everyone tells the interviewer you can't prioritise.
• Skipping the clarifying questions. Diving straight into solutions signals you didn't understand the prompt.
• Reciting the framework. "Now I'll do CASE-H, then PAT..." - this kills the conversation. Internalise the questions; don't perform them.
• Defending a weak idea instead of switching to a better one. Confidence is good. Stubbornness is bad.
• Forgetting to name trade-offs. Every product decision has them. Naming them is the senior signal.
• Ending without a strong summary. Question 7 is where the interviewer decides if you're hireable. Don't trail off.

Strong vs weak answers

A weak answer sounds like:

"I would add stories to LinkedIn. Stories are popular on Instagram and Snapchat, and they could drive engagement on LinkedIn too. I'd also add a dark mode and improve the notifications."

What's wrong: no user, no problem, no structure, features without justification.

A strong answer sounds like:

"I'd narrow this to new LinkedIn users in their first 30 days - specifically early-career professionals who joined to find a job. The problem for this user is that LinkedIn assumes you already have a network. New users see an empty feed, no connections, and no clear next step.

I'd propose three solutions: a guided 'connect with people from your school' flow, a personalised feed based on declared interests rather than connections, and AI-suggested introductions to people similar to them.

I'd start with the connect-with-school flow because it has the lowest activation friction and the highest signal of intent. The risk is that it feels gimmicky if the school overlap is weak - I'd validate the matching threshold before shipping. In three years, I'd want this to evolve into a full early-career mentorship layer that compounds the network effect for users who otherwise would have churned."

What's right: specific user, specific problem, multiple options, a defended choice, named risks, a three-year view.

How to prepare

Preparation for Product Sense is mostly about pattern recognition. The mental model is constant; what changes is the product. The more products you've thought about this way, the faster you can do it under pressure.

• Pick 10-15 products you use daily and write a one-page teardown of each. Walk through the 7 questions for each. Don't publish them. The point is to build the muscle.
• Study real PM teardowns. Cedric Chin, Lenny Rachitsky, Stratechery, Reforge. Read how senior PMs structure these arguments. Steal their structure, not their conclusions.
• Practice with a partner. Ideally a current or former PM. They'll catch the gaps in your structure that you can't see yourself.
• Time yourself. Most candidates either rush (15 minutes for a 30-minute question) or sprawl (50 minutes). Practice landing the answer in 25-30.
• Pre-build your "user library." Have 8-10 user segments ready - international students, freelance designers, gig workers, retirees, parents of teenagers, first-time homebuyers. When a prompt is vague, you can pick from this library quickly.
• Pre-build your "solution library." Have a mental list of solution archetypes - guided onboarding flows, AI-assisted recommendations, social discovery layers, trust-and-safety mechanisms. When you need to generate options under pressure, you'll have patterns to combine.