The PM role
What does a product manager do?
In this lesson4 sections
An unofficial lesson drawing on the methods in the The Product Manager Role chapter of Cracking the PM Interview by Gayle Laakmann McDowell and Jackie Bavaro (2014). PracHub is not affiliated with the authors or publisher. Buy the book
When an interviewer asks what a product manager does, they're testing whether you can show it in one product you shipped. They listen for the calls that were yours, how you got teams you don't manage to deliver, and whether it worked for users and the business.
We'll follow one feature from first evidence to months after launch. Maya is the PM for subscriptions at an invented meal-kit app with 80,000 active subscribers who get a weekly box and prepay monthly. About 6,000 cancel each month. Her team is four engineers, a designer and a shared data analyst. The feature lets a subscriber skip next week's box.
Give each call the evidence behind it
Maya's quarterly list had 12 candidates. Most arrived as one team's wish. Skipping arrived with subscriber evidence: about 900 support requests a month asked an agent to skip a week by hand, and in last quarter's churn survey, 31% of 1,400 people said they cancelled because they were away or still had food left.
She read all 434 of those survey comments and listened to ten support calls, so her case fit in two sentences:
"Support gets nine hundred requests a month to skip a week, and nearly a third of people who cancel say they were away or had food left. If they can skip in the app, some of them stay."
Skipping's real rival was grocery add-ons, the company's long-range bet, and the choice came down to proof against promise. Skipping had evidence now and would show up in cancellations within a quarter. Add-ons had none yet, needed a new catalog first, and would pay off in revenue years out. Maya argued they should wait, and they sit at the far end of the roadmap. Another of the 12, marketing's referral credits, goes second, since new subscribers are worth more once fewer leave.
Saying "later" is a call too. Near launch the commonest request was "let me pause for a month." Maya held it: skip data should first show how often people skip several weeks running. If the team builds every good idea before launch, the people who need skipping wait another quarter.
Answer Product Execution Walkthrough: Your Process From Concept to Release cold now and keep your notes for the end. Then try Bringing Customers Into Every Stage From Problem Discovery to Release and answer only the first part: how customers shaped which problem you chose.
Product Execution Walkthrough: Your Process From Concept to Release
A product manager execution question asking candidates to walk through how they took, or would take, a product from an early idea to its release.
Bringing Customers Into Every Stage From Problem Discovery to Release
A product manager interview question on how you involve customers from early problem discovery through building, beta and release.
Maya can build a clickable prototype with AI in an afternoon, and watching people use one changed two calls. She ran five sessions herself, each with one goal: "You're away next week. Make sure no box comes and you aren't charged." She watched without helping. Three of five looked under account settings first, so she took that to the designer and the skip toggle moved onto the home screen. Two couldn't tell whether they'd still pay, so she rewrote the confirmation to say the week would be credited.
Her designer owns the week picker: how it looks and how it responds to a tap. The rules beneath it are Maya's calls, and they rest on product sense: the judgment to say whether a product or plan will work for the people it's for, and why. Before any data, she saw that unlimited skipping would become a free pause. So a subscriber can skip at most three weeks in a row; a longer break is a pause, a different feature with a bigger risk to orders.
Mid-build, choosing several weeks on a calendar was taking far longer than estimated. Pressing engineers to go faster costs trust and rarely wins back the time, so Maya chose a list: version one shows the next four weeks with a skip toggle on each, and the calendar waits. Four weeks still covers the longest run of skips the rules allow; the cost is a plainer picker.
To practise this, try Partnering With Engineers Across the Product Lifecycle, From Idea to Release and answer only the build stage, around one moment when a plan changed mid-build.
You answer for the result, not the people
Maya doesn't write the code or draw the screens, and none of the teams skipping touches report to her: payments, the warehouse, support, finance, legal. She still owns the result.
The skip credit shows what that's like. A skipped week on a prepaid plan needs a credit on the next bill, and only the payments team can change credits. Maya couldn't assign that work. She could only make the case, so she raised it while writing the rules and brought the payments lead the support numbers.
The warehouse buys food on Thursday, so a skip, or undoing one, is allowed until noon on the Wednesday before delivery. Her own account caught a gap in that rule. She used her subscription on every internal build, and on a West Coast trip she skipped a week at 10am on a Wednesday. The app said "Skipped", and the box came anyway: the warehouse locks orders at noon Eastern, 9am where she stood. The rule looked complete on paper but never said whose noon. Now the app shows the cutoff in local time, and any bug that charges someone wrongly or ships a skipped box is fixed before launch.
Before launch, Maya named the failures that would hurt a subscriber and who would catch each. A wrong charge: on launch day the analyst and an engineer checked charges hourly. A box shipped after a skip: they watched skip volume hourly too, and the warehouse had a forecast of skipped boxes. Confusion about the credit: support, after a 20-minute briefing on what subscribers see and the three likeliest questions. What flips the switch that turns skipping off was her call, agreed with everyone two weeks out: a wrong charge or a shipped skipped box does, and confusion goes to support. When you tell it, that rule is "I" and the hourly checks are "we".
Where the PM's part ends differs between teams, so ask in interviews:
"When the PM and the designer disagree on a product rule, like how many weeks in a row someone can skip, who settles it?"
Try Defining the Product Manager Role: Accountability, Responsibilities and Scope aloud in your own words, then add one moment from your work. For a late launch call, try Ship-or-Delay Call on a Major Bug Two Days Before a Key Customer Release and say what you'd do in the first hour.
Defining the Product Manager Role: Accountability, Responsibilities and Scope
A behavioral product manager interview prompt asking the candidate to define the PM role in their own words.
Ship-or-Delay Call on a Major Bug Two Days Before a Key Customer Release
Cisco product manager execution question: two days before a release that a key customer depends on, the engineering manager reports a major bug.
Put the number against its target, miss included
Maya's ship question was whether skipping cuts cancellations without eating orders. Before the build, she turned both halves into numbers the team signed up to:
Main metric: monthly cancellations down 15%, from 6,000 to 5,100, within a quarter of full launch.
Guardrail: orders per active subscriber, now 4.3 a month, must not fall more than 2%.
Skipping can go too far, so set the guardrail where the boxes lost to extra skipping would outweigh the boxes from subscribers who stayed. Here a 2% fall costs about 6,900 boxes a month, while two months of 900 saved subscribers, at about 4.2 boxes each, bring back about 7,600.
She sized the test before trusting it. Skipping went behind a feature flag for a random 10% of subscribers, with the other 90% as control. Those 8,000 subscribers would cancel about 600 times a month without it, so a 15% cut (90 fewer) shows clearly within a month and a 3% cut wouldn't. An AI feature, say one suggesting weeks to skip, would also need a gate: a fixed set of test cases to pass before any subscriber sees it.
After a month, the test group was cancelling 11% less than control: a miss on the target, with orders per subscriber barely moved. Maya widened the rollout in steps, checking the guardrail each time, and kept a 5% holdout without skipping for a quarter to see if the effect lasted. Most test-group subscribers who cancelled anyway had never opened the skip screen, so to close the gap the cancel flow now offers a skip first.
Three months after full launch, subscribers with skipping were cancelling 13% less than the holdout: still short of 15%, but the effect had lasted. Explaining the miss falls to her, not to the engineers who built it. Maya kept skipping, and monthly pause, pending the skip data, is her next bet for the gap.
To practise the ship decision, try Ship or Kill: Turning A/B Test Results into a Launch Decision. For the speed-versus-quality call before it, try How a PM Decides When to Ship Fast and When to Harden a Release. Answer each aloud in five minutes before opening the hints.
Ship or Kill: Turning A/B Test Results into a Launch Decision
A product analytics question about making a launch decision from finished A/B test results.
How a PM Decides When to Ship Fast and When to Harden a Release
An execution question for a MongoDB product manager interview about balancing fast delivery against engineering robustness.
Turning this into "walk me through how you ship"
Asked how you ship, or what a PM does, tell this arc:
Use one real product, start to finish: a process with no product gives the interviewer nothing to probe.
Name each decision and the evidence behind it. "Three of five people looked in the wrong place, so I took that to our designer and the toggle moved to the home screen" beats "we did a usability study".
Say "I" for your calls and "we" for the team's work.
End with the number against its target, including a miss and what you did next.
Expect interruptions; they test whether the decisions were really yours. If you haven't shipped as a PM, use the closest thing you owned end to end and say what your role was. For the numbers, see metrics and diagnosing a drop.
Finally, tell your own story through PM Execution Story: Leading a Product from Concept Through Release in under four minutes and compare it with your notes from the start.