We used to send a take-home exercise to every designer who passed a first call. It was scoped at four hours. It reliably took people twelve, and the ones who could afford twelve unpaid hours were not a random sample of good designers.
What we do instead
A ninety minute paid session working on a real problem from our backlog, with two people from the team in the room. The candidate can ask questions, push back on the brief, and change direction, which is what the job actually involves and what a take-home cannot show.
Why it works better
A take-home measures the artefact. This measures the reasoning. We have hired people whose output in that session was rough and whose questions were excellent, and every one of them has been a good hire.
The cost
It is more expensive per candidate and it does not scale past about six people per role. We think that is the correct trade. A bad design hire costs a year.
We pay for the session at a standard contractor rate whether or not we make an offer. If you cannot do that, do not run this format.
