Product manager interview preparation, including where interviewers disagree

Most PM candidates prepare by collecting the one correct answer to each question shape. One framework for product sense, one story structure for the behavioral round. Then you get to the loop and the third interviewer cuts you off mid-framework, and the fourth one seems annoyed that you didn't have your framework ready.

This article covers the rounds you'll typically see, how to prepare for each, what to ask them back, and the six things senior interviewers openly disagree about. Read the disagreements section first, because it changes what you do in the other rounds.

The people quoted here are all on the record about how they personally interview: Nikhyl Singhal, who ran product groups at Meta and Google; Sarah Koo, senior director of product at Rippling; McKenzie Lock, a GM at Netflix; Sam Stone, VP of product at EvenUp; Bill Carr, who spent 15 years at Amazon and co-wrote Working Backwards; Karri Saarinen of Linear; Geoff Charles of Ramp; Katie Dill of Stripe; Jiaona Zhang of Webflow; Paul Adams of Intercom; Shishir Mehrotra of Coda; Keith Rabois of Khosla Ventures; Merci Grace, formerly of Slack; plus Lauren Rivera, who studies hiring at Northwestern, and Adam Grant on what the research says. Every video quote plays the exact moment it was said.

The rounds you'll probably get

Titles vary, but the loop at most product companies is some subset of these:

The take-home has quietly become standard. Netflix, Rippling, and EvenUp all use one, and Netflix runs it for every product role from individual contributor up through senior director. If you've been told take-homes went out of fashion, that information is a few years old.

Where experienced interviewers disagree

Here's the part generic guides skip. The people running these loops don't agree with each other, and the disagreements aren't small. If you prepare for one and walk into the other, you'll get a rejection you can't explain.

How much structure the interview itself has

The research here is pretty clear. Lauren Rivera, a sociologist at Northwestern's Kellogg School and one of the most cited researchers on hiring, spent an episode of Adam Grant's WorkLife taking apart the interview most companies still run.

Grant lays out the fix in the same episode: decide which skills the job needs, build questions around them, ask every candidate the same ones, and score the answers. He says the evidence suggests accuracy will often double or triple. He also names the reason unstructured interviews fail, which is that interviewers decide within about 90 seconds, and what they're reacting to in those 90 seconds is your voice, your face, and how much you remind them of themselves.

Some companies have gone all the way there. On a Skip panel of PM hiring leaders, Sarah Koo described Rippling's process, and it's the research recommendation taken literally.

Netflix does close to the opposite, on purpose. McKenzie Lock, a GM there, was answering the same question minutes earlier in the same conversation.

Sam Stone at EvenUp sits closer to Netflix, with different loops for different roles. So you have three companies with senior, experienced product leaders, and two incompatible philosophies about whether every candidate should get the same questions.

Amazon sits at the far end of the structured side, and has since 1999. Bill Carr, who spent 15 years there and co-wrote Working Backwards, describes a role built specifically to stop the hiring manager from going with their gut.

At 1:17:54 he gives the reason: the bar raiser's job in the debrief is to hold everyone to the written criteria, so nobody gets rejected for something like "they don't seem to want to work here enough." If you're interviewing at Amazon, every answer you give is being mapped to a named leadership principle, and you should know which ones.

What this means for you: in a standardized loop, the prompt is the same one they've given a hundred people, so your job is to be clearly better on the dimensions they score, and preparation pays off directly. In a loop the hiring manager designed, the interview follows your answers and the person may be scoring you against a picture in their head. There you need to stay flexible, and you need to give them a reason to like you, because a chunk of that decision is happening in their gut whether either of you likes it.

What your preparation signals to them

Follow the structured-interview argument to its end and you get a candidate who has practiced every likely question until the answers are smooth. Nikhyl Singhal, who spent years running product groups at Meta and Google, designed his favorite question specifically to break that.

His question is "what's something that everyone takes for granted that you think is essentially hogwash or inaccurate?" He explains at 17:19 that there's no way to answer it without being genuinely opinionated, because you can't work out what he wants to hear. That's the point.

Now hold that next to Rippling, where the case study prompt arrives 24 hours before the panel and the whole exercise measures what you did with those 24 hours. Both of these people are hiring good PMs, and they believe opposite things about what your preparation proves.

The way through is that they're testing different things. Singhal wants to know whether you have opinions of your own. The structured camp wants to know whether you can do the work. Rehearsing your stories until they're clear helps with both of those. What fails Singhal's test is rehearsing a position so hard that you can only defend it. So prepare the material, and go into every round with at least three things you actually believe about the product, the market, or the craft, including one that plenty of people in the industry would argue with.

How long you get to stay in questions

Every guide tells you to ask clarifying questions before you answer. That's right, but interviewers draw the line in very different places on how many is too many.

McKenzie Lock names the failure on one side. The candidates who trip up most, in her experience, are the ones who never ask whether it's the right problem at all and just start answering. At 31:32 she calls that a big red flag.

Shishir Mehrotra, CEO of Coda, built a question that punishes the other side. He tells you a group of scientists have invented a teleportation device and hired you to bring it to market, then lets you ask as many questions as you want before cutting you off.

He's watching which two you keep. His favorite answer was a PM who asked whether it's safe enough for humans and whether the cost is mostly capex or opex, then built a plan off that grid. Ask about the size of the device and he'll ask you what decision that would help you make.

Lock also warns about candidates who ask so many questions that the panel ends up driving the conversation. So the same person wants you to question the problem and wants you to stop questioning it in time. That's a real skill, and it's the one to practice: ask the two or three questions that would change your answer, say why they'd change it, then commit.

Whether "what would you do at our company" is a good question at all

You will almost certainly be asked some version of "if you ran the company you're at now, what would you change?" or "what would you do about our product?" Two experienced interviewers have opposite views on whether that question is worth anything.

Keith Rabois, who was COO of Square and an executive at PayPal and LinkedIn and now invests at Khosla, uses it as his main question for senior hires.

Shishir Mehrotra calls the same shape of question a bad way to judge people, for a specific reason: the interviewer has spent years on the problem and you've spent twenty minutes, so all you can really do is guess their answer.

That's also why his teleportation question is about a product that doesn't exist. Nobody has a home court advantage on it.

You don't get to pick which one you're facing, so prepare for the Rabois version, since it's the one that can go badly. Have a real answer for what you'd change at your current company, and have the follow-up ready too, which is why you didn't get it done. "I never raised it" is a bad answer. "I raised it, here's how, here's why it lost to the other priority, and here's what I'd try differently" is the answer he's fishing for.

Whether where you've worked counts

This one matters most if your resume doesn't have a famous logo on it. Sarah Koo weights past employers heavily, and she knows it's contested.

Sixteen minutes later, McKenzie Lock was asked whether Netflix mostly hires out of Google, Meta, and Amazon, and gave a flat no.

Koo hands you the workaround herself at 22:25: trajectory. Four promotions in six years at a slow company still stands out to her, because it shows you did well in a place where doing well wasn't automatic. If your logos are unknown, put the trajectory and the outcomes at the front of your story, and expect a back channel. The positioning playbook covers how to build that framing before you apply.

Whether you should work for free to get the job

Take-homes and work trials are now common enough that you'll face this. Karri Saarinen, co-founder and CEO of Linear, ends every hiring process with a work trial, and Linear pays for it.

Merci Grace, formerly head of growth at Slack, ran a project too, on the candidate's own time and unpaid, and she treats an objection to that as a signal in itself.

So one founder treats the trial as contract work and pays for it, and one product leader reads your willingness to spend an unpaid evening as evidence you want the job. Practically: ask what the time budget is, ask whether the trial is paid, and decide how much unpaid work you'll do across your whole search rather than company by company. Linear's trial also runs both ways, and Saarinen says at 1:04:17 that it's genuinely hard to tell how a startup operates from interviews alone, which is worth remembering when you're the one deciding.

The same story, graded on different things

Here's the practical consequence of all of the above. You'll tell the same career story four times in one loop and be graded on something different each time, because these people are listening for different signals. From one Lenny's Podcast compilation of interview questions from more than 100 guests:

Paul Adams saves his best question for your references, and it's phrased so they can't be vague.

One story, told well, can cover Charles's agency question, Dill's taste question, and Zhang's ambiguity question, but only if you built it to. Pick two or three projects and write each one out so it carries all three: what made it hard, what you chose and why you thought that choice was good, and how you moved from confusion to a plan. Then answer whichever version of the question you get. Adams's reference question is a reminder to name your real weakness before someone else does, and to know which former manager will pick up the phone.

What startups screen for that big companies don't

The stage of the company changes the weighting more than any framework does. Lenny asked Geoff Charles what Ramp looks for that other companies undervalue.

He says outright that Ramp overweights those two things relative to experience, because experience matters a lot less at a business shaped like theirs. He also treats your reason for leaving as a signal, and the one he likes is a candidate saying things got too slow and too bureaucratic. That's close to the opposite of the answer a big-company interviewer wants, where the safe framing is that you'd grown as far as the scope allowed.

Sam Stone's version of this at EvenUp is that they look for people who have worked somewhere with a similar stage, growth rate, and culture, and that they've been burned hiring people from the same vertical who arrive convinced they already know how it's done. So domain expertise carries less weight than you'd think, and having survived a comparable pace carries more.

A few practical differences that follow from stage:

How to work out which kind of interview you're in

You can find out before you walk in, and you should. Most of it comes from the recruiter, who will tell you if you ask.

If a company won't tell you its format, treat that as information about the company.

How to prepare for each round

What to ask them

The questions you ask get scored, and they're also the only way you find out whether you want the job. Phyl Terry, who wrote Never Search Alone, has candidates draft their own version of the job description before they interview.

Beyond that, a few that consistently earn a real answer:

Frequently asked questions

How long should I prepare for a PM loop?

Three to four weeks if you're currently working, and most of that on stories and product sense reps out loud. If your onsite is tomorrow, the 24-hour prep playbook covers what to cut.

Should I use a named framework like CIRCLES or STAR?

Use a structure, and don't announce its name. Interviewers at large companies have heard every acronym hundreds of times, and naming one earns you nothing. The bigger risk is that a framework makes you rigid, which is exactly what gets punished in a loop where the interviewer follows your answers rather than a script.

What if I get rejected without feedback?

Given how differently these companies run their processes, a rejection often says more about fit with one interviewer's model than about your ability. Ask anyway, since some recruiters share notes. The post-rejection playbook covers what to do with the answer.

Do I need a well-known company on my resume?

It helps with Koo and it doesn't decide anything with Lock, who says most of her hires came from companies you haven't heard of. Charles weights hunger and depth of thinking above experience entirely. So the answer varies by who's reading, and the thing that travels everywhere is evidence you've done the work, which is easier to show than a logo is to acquire.

How much do metrics in my answers matter?

Numbers make a story credible, so bring them. Then expect the follow-ups to be about why you made the call, because that's where every interviewer quoted here spends their time. If you can only recite the result, you'll get found out in the second question.

Are take-homes worth the time?

They're now most of the signal at several companies, so you rarely have a choice. Ask how much time they expect. EvenUp tells candidates not to spend more than four hours and says the correlation between hours spent and how well people do has been getting weaker.

Your PM interview prep checklist