Design interviews are not one conversation with different questions. They're four or five separate exercises, each testing something different — and most candidates prepare as if it were all one thing.
That's why a designer with a beautiful portfolio can sail through the walkthrough and then fall apart in the critique round. Different stage, different skill, no warning.
So this one is organised by stage rather than by topic. Here's the map first.
| Stage | What it's actually testing | Where candidates lose it |
|---|---|---|
| Portfolio walkthrough | Whether you can tell a story with stakes, and separate your work from the team's | Narrating process chronologically instead of leading with the problem |
| App critique | Whether you have judgment, not just taste | Listing what looks dated instead of asking who the user is |
| Design challenge | How you think when you don't know the answer | Jumping to solutions before scoping the problem |
| Craft and systems | Whether you can ship inside real constraints | Talking about ideal process that no team actually has |
| Collaboration | How you behave when someone disagrees with you | Stories where the designer is always right |
The portfolio walkthrough
Usually the first round, and the one that sets your level for everything after.
- Walk me through this project.
- What was your specific contribution, and what did other people do?
- How did you know whether the design worked?
- What would you do differently if you started it again?
- Why this solution, out of everything you explored?
Saying it out loud: question 1 is the single most-rehearsed and most-fumbled moment of the loop. The failure mode is chronology — starting at the kickoff and walking forward through every workshop. Lead with the stakes instead: "Only 30% of new users ever created their first project. I owned onboarding, working with one PM and two engineers. The interesting part was that the drop-off wasn't where the funnel said it was." Now the interviewer wants to know what happened. Chronology gets you a polite nod at minute six.
The rule of thumb: if your first sentence could describe any project at any company, it's the wrong first sentence.
Opening with process
“This project ran for four months and involved twelve stakeholders.”
Says nothing about whether the work mattered, or which parts were yours. The interviewer now has to wait — politely — for the interesting bit.
Opening with stakes
“Checkout conversion dropped 18% after a release, and I owned the fix.”
Problem, scale and your role in one sentence. Everything after it has somewhere to land, because the interviewer already knows what was at risk.
The app critique
Middle rounds. Often unannounced — they'll just hand you their product.
- Pick an app you use every day and critique it.
- Here's our product. What would you change first?
- Everything on this screen has problems. How do you decide what to fix first?
- What's a design decision in a well-known product that you think is wrong, and why?
Saying it out loud: the trap in question 7 is that it sounds like an invitation to be impressive, so candidates start listing flaws. Judgment means asking before judging: "Before I critique it — who's the primary user here, and what's the one metric this screen is meant to move? Because if this is for first-time users, the thing I'd change first is completely different than if it's for power users." Then critique. Interviewers are watching whether you can tell the difference between a preference and a problem.
The design challenge
The whiteboard or take-home round. Senior loops weight this heavily.
- Design a way for people to do a specific task on a specific platform.
- What are the edge cases — empty state, error, offline, first run?
- When do you follow the platform pattern, and when do you invent something new?
- How would you validate this before anyone builds it?
- You have two weeks and no researcher. What do you actually do?
Saying it out loud: the challenge round is graded on the first three minutes, not the drawing. Spend them out loud on constraints: "Let me make sure I'm solving the right thing. Who is this for, what are they trying to get done, and what does success look like for the business? And is there a constraint I should know about — platform, timeline, existing design system?" Candidates who draw immediately look fast. Candidates who scope first look senior. Only one of those gets the offer.
Why this works across all three rounds so far: the portfolio, the critique and the challenge are the same test in three costumes — can you separate the problem from the solution, out loud, before you commit to an answer? Interviewers change the format precisely to see whether that habit is real or rehearsed.
Craft, systems and handoff
Asked at every level. This is where interviewers find out if you've shipped.
- How do you work with a design system — and what do you do when a component doesn't fit?
- What do you actually check for accessibility?
- How do you hand off to engineers, and what do you do when something ships slightly wrong?
Saying it out loud: question 15 has a trap in the second half. "I'd create a new component" is the wrong instinct, and "I'd never deviate" is the wrong answer too. What they want is a decision process: "First I check whether an existing component can do it with the variants we already have — most of the time the answer is yes and I was being precious. If it genuinely can't, I'd take it to the design system owner rather than forking it locally, because one-off components are how systems die."
Collaboration and disagreement
Behavioural round. Weighted more heavily for Senior and above than most designers expect.
- Tell me about a time an engineer told you your design wasn't feasible.
- How do you handle a stakeholder pushing for something you believe is wrong?
- Tell me about feedback that genuinely changed your design.
Saying it out loud: every one of these has the same failure mode — the story where the designer was right all along and eventually everyone realised it. That's not what's being tested. Question 20 exists specifically to find out whether you can be moved by evidence: "I was convinced the filter should be persistent. An engineer pointed out it would double our query load on the busiest screen, and honestly the usage data didn't justify it. We shipped it collapsed by default and nobody complained." Being wrong, cheaply, in front of an interviewer is a strong signal.
The part no portfolio can carry
Every stage above is scored on something a beautifully crafted portfolio cannot demonstrate: how you sound while being questioned about your own decisions. The critique round in particular is designed to push back on you — and the moment where you either defend a weak point or gracefully concede is worth more than the critique itself.
If that moment tends to rattle you, it's worth practising separately from the design work. We wrote about how to reset mid-interview in How to recover after a bad answer.
Practical target: record yourself walking through one case study for three minutes, out loud, with no slides. If you can't make the problem and your role clear in the first thirty seconds without visuals, the portfolio is doing work your narration should be doing — and in the interview, narration is what gets graded.
