Last updated: October 13, 2026
Good product designer interview questions for 2026 ask a candidate to find the funnel step worth redesigning, read an A/B test honestly, and design for an AI feature that is sometimes wrong. Portfolio polish tells you about taste. These questions tell you whether the person will move a number.
Full disclosure on where I sit. I have never shipped a Figma file in my life. Most of my week at KORE1 is commercial, which puts me on the debrief calls after a design hire goes sideways, and over the past couple of years those calls have started to sound alike. The portfolio was gorgeous. The case studies were tight. Six months in, nobody could point to a metric the new designer had changed.
So our creative recruiters and I started writing down what the panel should have asked. This is that list, built for the title that carries business context, not just craft.
We earn a fee when a designer we introduce through our product designer staffing desk accepts an offer, so read with a little suspicion. Nothing here requires us.
Some things are deliberately missing. The live Figma file walkthrough, the two-sentence framing exercise, and the stakeholder scenario are on that staffing page. Portfolio red flags, the paid design exercise, and the product-versus-UX title fork belong to our product designer hiring guide. Pay is the product designer salary guide‘s job. And when the seat is mostly research and flows, the three-round loop in our UX designer interview questions is a better fit than anything below.
Hand Them a Funnel Before You Hand Them Figma
A payroll software company in Columbus, Ohio, hired a product designer last year with the best portfolio I have seen come through our desk. Fintech dashboards, a design system with 140 components, clean motion work. She spent her first quarter rebuilding the settings pages, and they did look better. Meanwhile 47% of new accounts were stalling at the bank-connection step, which nobody had shown her, and she never thought to ask for the funnel. The VP of product called us in month seven.
Nobody on that panel had put a number in front of her. Why would they? She had shipped at two well-known companies.
We now suggest a one-page funnel sheet. Twenty-five minutes, paper or a whiteboard, no laptop. The fictional product we use is a scheduling app for dental offices, and I like it because every hiring manager instantly pictures a front desk. Of its 10,000 monthly signups, 8,200 verify an email, 4,100 connect a calendar, 2,870 invite a teammate, and 2,296 book a first appointment inside seven days, so activation sits near 23%.
Give the candidate those five numbers and the five problems below, one at a time. Arithmetic isn’t what gets scored. Watch whether they reach for the end metric or stop at the step metric, and whether they ask what the users were actually doing before they suggest a single screen.
| The prompt | Where a strong answer lands | The tell in a weak one |
|---|---|---|
| Which step would you study first? | Calendar connection. Half of verified users leave there, 4,100 people a month, the biggest absolute loss in the funnel. | Picks the step whose screen looks oldest |
| The PM wants to redesign email verification. Lifting it from 82% to 90% adds how many activated users a month? | About 224. Lifting calendar connection from 50% to 60% adds about 459, roughly twice as many, for one step’s worth of work. | Agrees with the PM without running either number |
| A test on the calendar step shows 53.0% against 50.0%, with 1,000 users per arm. Ship it? | Not yet. A three-point gap at that sample size appears by chance nearly one time in five. Detecting it reliably takes around 4,350 users per arm. | “Three points up, that’s a win” |
| A variant makes the calendar optional. Step three jumps to 75%, but first bookings fall to 55% of inviters. Better or worse? | Run it to the end. About 2,368 activated against 2,296, so a wash, and those offices may never sync a calendar. | Reports the 75% and stops |
| What do you want to see before you open a design tool? | Ten session recordings at step three, a split by Google Workspace versus Microsoft 365 offices, and two calls with practice managers who quit there. | Starts describing a new layout |
That last row is where the best candidates separate themselves, and it usually happens without prompting. Someone who has shipped onboarding in a B2B product will guess out loud that the calendar step probably triggers an admin consent screen at offices running Google Workspace or Microsoft 365, and that the front-desk person who signed up can’t approve it. That guess is worth more than a redesign, because the fix might be an “invite your IT person” path rather than new pixels.
Columbus rehired in the spring. The designer they chose looked at the funnel for maybe a minute, then asked what share of signups came in on a phone. Nobody in the room knew. Good question.
Questions About What Happened After Launch
Product designers get hired to change outcomes, so ask about outcomes. Most portfolios describe the problem and the final screens and skip the part where the thing went live.
- “Pick one project. What number was supposed to move, and what did it do in the eight weeks after release?” Plenty of designers never saw the number. That is common, and an honest “I never got the dashboard, so here is what I asked for” is fine. Inventing a figure is not.
- “How do you know your change caused it?” Listen for a holdout group, a staged rollout, or at least a candid admission that a marketing push landed the same week. A 12% lift that coincided with a pricing change is not a design win, and senior people know it.
- Ask for something they recommended not building. Product designers with real seniority have talked a team out of a feature, usually with a five-user test or a support-ticket count, and they remember the meeting.
- What did they measure that turned out to be the wrong measure? One candidate told our recruiter about optimizing time-on-page for a help center and realizing, two months later, that longer visits meant people could not find the answer. That story got her the offer.
Short answers are fine here. Vague ones aren’t.
Designing for an AI Feature That Is Wrong Some of the Time
This is the newest question on the list, and in 2026 it’s the one most likely to surprise a candidate who has spent the last three years on dashboards and checkout flows. Nearly every product roadmap we see now has a generative feature somewhere on it. Somebody has to decide what the user sees when the model is confident and incorrect.
Use a concrete prompt. “Our invoice tool reads a supplier PDF and fills in vendor, due date, line items, and total. It gets every field right on 92 of 100 invoices. Design the review screen.”
Weak answers add a sparkle icon, a disclaimer in gray text, and a big green Approve button. Strong answers start somewhere else entirely. They ask which fields hurt when they’re wrong, and they treat a wrong total on a $48,000 invoice differently from a misspelled vendor name. They show the source line next to each extracted value so checking takes two seconds instead of twenty. They make correcting a field cheaper than retyping it, and they log those corrections, because every edit is free training signal and a free quality metric. Some will suggest the tool say nothing at all about confidence on low-risk fields and force a confirmation on the dollar amount only. That’s a sophisticated answer.
If the candidate cites prior work, even better. Microsoft’s Guidelines for Human-AI Interaction, published with a CHI 2019 paper, organize their advice by moment, including what a system should do when it is inevitably wrong. Google’s People + AI Guidebook covers explainability and error states in the same spirit. A designer who has read either one tends to talk about trust calibration rather than about “AI magic.”
A second AI question works well for senior seats. Hand over a flow generated by a prompt-to-UI tool such as Figma Make and ask the candidate to critique it in ten minutes. These tools produce plausible screens fast. The interesting part is whether the candidate spots the missing empty state, the error path nobody prompted for, and the component that quietly ignores your design system.
The Cancel Flow Question
Ask this one with your head of finance in mind. “Finance wants three screens and a retention offer between the Cancel button and an actual cancellation. Design it.”
No answer is correct here. That’s deliberate. You learn how the candidate handles a business ask that pushes against users. The weak versions go one of two ways. Some candidates build the obstacle course without a word of objection. Others refuse on principle and offer nothing to the business. Neither person is going to survive long in a company that has a revenue target.
The answer we like sounds roughly like this. One honest offer, maybe a pause or a cheaper plan, on a single screen with an obvious way out. A reason picker that feeds the product team real churn data. And a sentence about risk. The Federal Trade Commission’s 2022 staff report, Bringing Dark Patterns to Light, describes cancellation obstacles as one of the tactics regulators watch. A product designer who can say that calmly to a CFO, and still hand back a design that keeps some customers, is worth a lot.
One Question About the Design System
“You need a component the library doesn’t have. Do you detach an instance and build it, or propose a new variant?”
It sounds like trivia. It isn’t. The answer reveals whether someone treats the system as a shared contract with engineering or as a suggestion. Good candidates count. A one-off on an internal admin page, detach and leave a note. Once three teams need the same thing, it goes to whoever owns the system, with the use cases attached. Candidates who always detach have usually never worked somewhere with a maintained library. Candidates who always wait for the system team have usually never hit a deadline.
Who Sits in the Room, and for How Long
Put a product manager and an engineer on the panel, not just designers. The funnel sheet goes best with the PM running it, because the PM will notice instantly whether the candidate’s instincts match how decisions get made on your team. The engineer should own the AI review-screen question, since the cheapest correction pattern is usually an implementation question in disguise.
Two sittings covers it. The funnel sheet and the after-launch questions take about an hour. The AI question, the cancel flow, and the design system question fit in a second hour with the hiring manager. Score product judgment, craft, and communication as three separate marks. A candidate who aced the funnel but couldn’t defend a cancel screen to a skeptical PM is a different hire than one with the opposite profile, and averaging hides that.
Don’t let it drag. Federal wage data puts web and digital interface designers, the occupation where most digital product designers get counted, at a $104,000 median in May 2025 across 113,330 jobs. The BLS Occupational Outlook Handbook expects the wider web developer and digital designer group to grow 5% between 2025 and 2035, with roughly 13,600 openings a year. Not explosive. Still, designers who handle the funnel prompts well are a thin slice of those 113,330, and they don’t stay available long. If the offer band is what’s slowing you down, the salary benchmark assistant returns a range in about a minute.
Twelve months after start, 92% of KORE1’s direct hire placements are still with the employer that hired them. Most of that is decided by how the role gets scoped before a panel ever meets.
Things Hiring Teams Ask Us About These Questions
Is the funnel sheet fair to a designer who never had analytics access?
Yes, as long as you score the reasoning and not familiarity with a particular dashboard, because finding the biggest absolute drop and following it to the end metric needs no tool at all.
Plenty of agency designers and designers at early startups never got a Mixpanel or Amplitude login. Tell them up front that it’s fine. The ones who can think this way figure it out in the room, usually by the second prompt.
How is this different from interviewing a UX designer?
Product designer questions add business ownership, meaning a metric, a tradeoff with revenue, and a call about what not to build, on top of the research and flow skills a UX loop already covers.
If the person will mostly run research and shape flows inside someone else’s roadmap, use the UX loop instead. Asking a strong UX specialist to defend a retention offer to finance tests the wrong job.
Do we still need a take-home exercise?
Usually not, since the funnel sheet and the AI review screen give you a working sample in about forty minutes of live time without eating anyone’s weekend.
Should junior candidates get the same questions?
Give juniors the first two funnel prompts and the design system question, and skip the cancel flow and the Figma Make critique, which reward experience they have not had time to collect.
For a junior seat, a clean “I’d want to see the recordings first” counts as a strong answer. You are hiring for the instinct to ask, not for the right answer on day one. One more thing. A junior who confidently invents a post-launch metric is a bigger risk than one who admits they never saw it.
What if the candidate gets the A/B question wrong?
Treat it as information rather than a failure, because plenty of good designers have never been taught sample size, and how they react when you explain it matters more than the first answer.
Watch for curiosity. “Oh, so how many users would we need?” is a good sign. Defensiveness about a statistics question during an interview is usually what shows up later in a design review, too.
Can these questions work for a contract product designer?
Trim the set to the funnel sheet and the AI question for a contract seat, since a contractor typically joins mid-project and needs judgment on a defined problem more than a long-term vision.
We place a lot of product designers through contract staffing for launches and redesigns that run eight to twenty-six weeks. A single one-hour sitting is usually enough, and a fast offer matters more on contract than anywhere else.
Before You Book the First Panel
Run the funnel sheet past your own strongest designer or PM first and time it. If they need more than twenty-five minutes, cut a prompt. If they finish in eight, swap in real numbers from your own product, which also tells you quickly whether your team has those numbers handy. Many don’t, and that’s useful to know before you hire someone to move them. Put that number in the posting too, the way this product designer job posting template does.
Most senior product designer seats we fill turn into direct hire placements, because the context a designer builds about your users compounds every quarter they stay. If you want a shortlist of designers who have already owned a metric, or just a second opinion on your loop, talk to our creative recruiting team. And if your panel already asks questions like these, we’ll tell you that on the first call.
