Product design questions

A product design question in a PM interview ("How would you design...?") tests whether you can choose a goal and a user before you name a single feature. Cracking the PM Interview puts design, improvement and favorite-product questions in one family. We'd put it this way: none of the three can be judged until you've said what the goal is. When a candidate jumps straight to screens, an interviewer can't tell whether they found the right thing to build or just the first thing they thought of.

We'll carry one prompt all the way through: "How would you design a laundry-room booking system for a large apartment building?" It's good practice because the money and the use sit with different people. The owner pays. The people using it are tenants with baskets and the contractor who repairs the machines.

What the prompt leaves undecided

Read the prompt again and circle every word that could go more than one way. "System" might mean an app, a text message or a sheet of paper taped to the door. For each one that would change your design, state what you'll assume and let the interviewer correct you. A default costs them one word to fix, where an open question hands them your work.

If a prompt seems to contain nothing at all, our note on vague product prompts goes further. Out loud:

"I'll assume one landlord, 300 apartments, one laundry room with 10 washers and 10 dryers, and machines we can't connect to software in the first year. Tell me if you pictured something else."

For practice on this step alone, take Design a Camera-Equipped Smart Mirror: Pick the User, Use Case and First Version, write your defaults, then stop.

What winning looks like before you name a feature

Ask what the owner would call better a year from now: fewer angry calls to the manager, tenants who renew their leases, and no pressure to buy more machines. Renewals move too slowly to judge a laundry system, so I'd pick something the manager can count every month: laundry complaints.

A little arithmetic first tells me what kind of problem this is. These are all my assumptions:

• The room is open 7am to 10pm, which is 15 hours, and one washer handles one load an hour.

• 10 washers × 15 hours × 7 days = 1,050 wash slots a week.

• 300 apartments × 2 loads a week = 600 loads. 600 ÷ 1,050 ≈ 57% of slots.

Across the week, there are machines to spare. Now look at the crowd. Say half of all loads, 300, want Saturday or Sunday between 8am and 6pm. That window holds 10 washers × 10 hours × 2 days = 200 slots, so 100 loads don't fit. The other 850 slots in the week have only the remaining 300 loads to hold.

That changes what winning means. Clearing the weekend with machines alone would take 300 loads ÷ 20 hours = 15 washers, five more than today, and they'd sit idle most of the week. The building needs some loads to move, and it needs people to find out before they carry a basket downstairs.

"Winning is tenants who can count on doing laundry when they planned to. I'd measure it by laundry complaints to the manager each month, and I'd aim to halve them in three months without total loads falling."

Halving is my judgment call. The last clause is a guardrail, a number that warns you of harm elsewhere. If fed-up tenants start using a laundromat, complaints drop, and so does what the owner earns from the machines.

To practice, try Redesign the DMV Experience: write one sentence on what winning means for whoever runs it, and stop before you think of a feature.

How to pick a user

List everyone the system touches and what each one wants from it:

• The owner pays, and wants fewer complaints and no new machines.

• The building manager takes the complaints and would run the system day to day.

• The repair contractor wants to know which machine broke, and how, before coming out.

• Working parents do many loads and can mostly do them on weekends.

• Tenants with flexible hours, such as retirees and shift workers, could go at quiet times but have no reason to.

• Tenants without a smartphone can't use anything that lives only in an app.

The person who pays may never open the product. Forget the owner and you've designed something nobody buys. Design only for the owner and you've built a complaints dashboard while the basement stays as it was.

Then choose one user for depth, out loud, and say who you dropped. I ask two things: who runs into the crowded hours most often, and who, once served, would do the most for the goal?

"I'll design for a working parent doing five loads a week, mostly on weekends. They hit the crowded hours most, so I'd guess they're behind a good share of the complaints. I'm dropping tenants with flexible hours for now, and I'll come back to them, since they're the ones who could move. The contractor gets one small feature, because a broken machine hurts everyone."

Naming who you dropped is what makes the choice real. "We'll serve all tenants" commits you to nothing, and an interviewer will likely keep pushing until you choose.

For either question below, write the list of people and the one you chose. Nothing else.

One person's real week in the laundry room

Tell that person's week as a story, with times and floors in it. A story turns up failures that a neat list of tasks misses.

Dana works full time and has two small kids. Saturday at 9, during nap time, she carries a basket down from the sixth floor. Every washer is running, so she goes back up. At 10:30 she tries again. One washer has stopped, with someone else's wet clothes still in it. She waits ten minutes, then moves a stranger's laundry onto the folding table, feeling bad about it. An hour later her wash is done and every dryer is full, so her clothes sit wet in the basket. On Sunday she walks down to a row of full washers again. Meanwhile one washer has worn a handwritten "broken" note for two weeks, with no date on it.

Now write down each failure and roughly what it costs her. The figures are my guesses, and in the room I'd say so.

Minutes miss something, though. Handling a stranger's wet underwear costs ten minutes, and it's the story most likely to reach the manager, and complaints are my goal. So I move it above the dryers. My final order is washer taken, someone else's load, dryers full, broken machine.

Try this on Design a New ATM Tailored to Airport Travelers: write one person's story, rank the failures, and hold back every idea for now.

Ideas, each tied to a failure

Only now do ideas come out, each beside the failure it fixes. An idea with no failure next to it goes.

Now cut. Here is what I'd build first:

• Weekend booking. Reserve a wash hour by phone, or on a sheet the manager keeps for tenants without one, with a dryer held for the hour after. Weekends only, since the week has room. A slot unused after 15 minutes opens to anyone.

• The 15-minute rule, with bins and a reminder text. Mostly a sign and some plastic bins.

• The QR code. It's the contractor's one feature and costs little.

Here is what I gave up, and why:

• Live machine status needs a sensor on each machine, and I assumed none in the first year. Booking removes most of the reason to check anyway.

• More washers cost the owner the most and sit idle on weekdays.

• Off-peak discounts could pull flexible tenants out of the weekend, but I took price as fixed. If booking doesn't move enough loads, this is my next test.

Booking has a cost too. 300 loads still want the 200 daytime slots on Saturday and Sunday, so 100 of them, one in three, will find the daytime full. They can still book early or late on the weekend, when the room has another 100 slots (7am to 8am and 6pm to 10pm: 5 hours × 10 washers × 2 days), or on a weekday. That's a smaller problem than today's: an hour they didn't want, instead of a wasted trip downstairs. Still, someone who used to walk down on a whim now has to plan ahead, and I'd say that before the interviewer does. Our article on explaining product decisions and trade-offs goes further on defending a cut.

Run Design a New Credit Card up to this point: ideas tied to failures, one cut, one thing you gave up.

End on a recommendation the interviewer can repeat

After you leave, your interviewer walks down the hall and a colleague asks what you proposed. Hand them the sentence they'll use. Put four things in it: who it's for, what gets built first, what got left out, and the number that would tell you it failed.

"For a 300-apartment building, I'd start with weekend booking for working parents: reserve a wash hour from your phone, with a dryer held for the hour after and a 15-minute rule for loads left behind. I'd add a QR code on every machine so breakdowns reach the contractor directly. I'm leaving out sensors and new machines. If laundry complaints haven't halved in three months, or total loads have fallen, I'd pull it."

Then stop and let them respond.

If the prompt hands you an existing product instead, start from where its users give up, as in product improvement questions.

To practice, work through Product Design: A Guided Tax-Filing Product Like TurboTax for Individual Filers and say your final recommendation aloud in under 30 seconds.