What we believe about hiring
These nine positions explain why the product is shaped the way it is, and why it refuses to do some obvious things.
- 01
Presentation and judgment are different skills
The design and product interview has always rewarded whoever narrates their work most comfortably. That trait is only loosely related to the ability to make good decisions, and the gap is widest at senior level, where the calmest presenter is often the one who was shielded from the hard calls.
- 02
The brief always arrives wrong
Nobody is handed a well-posed problem. Requests arrive as solutions, with the answer already chosen by someone who isn't doing the work. The first act of the job is noticing. Any assessment that hands over a clean brief is measuring the wrong half of the role.
- 03
What happens when the plan breaks is the job
Every case study ends well, because it's written after the outcome is known. The abandoned direction gets tidied into an exploration and the constraint that ruined the quarter disappears entirely. The part that predicts seniority is exactly the part the format deletes.
- 04
An assessment owes the candidate their time back
The industry treats a candidate's hours as free. So we ask for fifteen minutes, tell them the criteria before they start, and make it count across companies: if someone else asks for the same assessment, they don't sit it again.
- 05
Unpaid weekend work filters for privilege
A twelve-hour take-home selects the candidate with the fewest competing offers and the fewest caring responsibilities. That filter is invisible in your funnel (the people it removes leave no trace), so teams mistake their shortlist for the market.
- 06
Banning AI is a lie you tell yourself
Detection is unreliable and its false positives land hardest on non-native speakers. The better response is to design tasks where a model helps less: context that exists only inside the exercise, constraints that move, and reasoning that has to be defended out loud.
- 07
Don't act on a score you can't audit
A percentile with nothing behind it asks you to outsource a decision about a person's career to a black box. Every recommendation we publish says what drove it, and any candidate can ask a person to review it, so you can read it and disagree with us.
- 08
Depth beats breadth, and we chose depth
We could cover forty roles tomorrow and be a worse version of something that already exists. Six roles assessed properly is more useful than four hundred assessed shallowly, and pretending otherwise would be a growth decision dressed as a product one.
- 09
Being wrong in public is part of the method
The rubric is versioned because we expect to change it, and changing it breaks comparability, so we announce every change instead of editing in silence. An instrument that never admits error is just a brand.
If you disagree with any of these, we’d like to hear it. Several of them exist because somebody argued us out of a previous position.
Related reading
What AI broke, in detail
10 min
How to tell if a design portfolio was generated with AI
You usually can't tell from the artefact itself, and detection tools don't work reliably. What you can do is look for missing specifics (real constraints, abandoned directions, disagreements, numbers that are inconveniently precise) and then ask one interview question that only someone who was in the room could answer.
Read it8 min
Hiring designers in the age of AI
Generative AI collapsed the cost of producing the artefacts hiring used as proxies: screens, case studies, cover letters, application volume. It didn't change what makes someone good at the job. The practical response is to stop assessing artefacts produced in private and start observing reasoning under conditions you control.
Read it