Portfolio

What a portfolio actually is

A portfolio is work the candidate already made, on their own schedule, usually before they ever saw this job posting: a set of design files, a GitHub repo, published articles, launched campaigns, a case-study deck from a past employer. Nobody at your company briefed it, scoped it, or set a deadline for it.

That’s the opposite of a work sample: a task built by the employer, for this hiring round, using inputs the employer chooses. A portfolio wasn’t designed for you at all: it existed before you applied, and the candidate picked which pieces to show you. Review one as if it were the other and the mistake runs in a predictable direction: grading unstandardized, self-selected work as if a controlled test had produced it.

How a good review actually works

Pick one or two pieces and ask the candidate to narrate them: what the client or manager actually asked for, what got cut, why. A candidate presenting a shipped rebrand talks about the final screens; a candidate narrating it talks about the three directions they rejected and the one constraint that decided the outcome.

Nielsen Norman Group’s 2019 survey of 204 UX hiring professionals found the same expectation from the other side of the table: respondents said they want to see the messy process and the research behind the work, instead of only the finished, polished screens. What’s worth checking, across whatever the candidate brings:

• Range. Does the work span more than one context or constraint, or is it one good idea shown five times?
• Judgment. Can they name a decision they’d defend and one they’d change now? Silence on both is the tell.
• Relevance. Does anything resemble the actual role you’re hiring for, or is this a generic best-of reel built for a different kind of job?

Where portfolio review breaks

The most common failure is grading the artifact instead of the reasoning behind it. A beautifully typeset case study can hide a candidate who copied a template’s structure; a rough, badly formatted one can hide someone who made every call themselves. Polish is easy to fake with a good template. Explaining a real decision, live, in conversation, is much harder to fake.

The second failure is treating the portfolio as the whole assessment. It can’t be: nothing about it is standardized. Candidates chose what to include, spent wildly different amounts of time polishing it, and got different amounts of help along the way. The US Office of Personnel Management describes a work sample test as an evaluation run under realistic, standardized conditions. A portfolio is neither, by design, and scoring it as if it were invites the same loose judgment a structured process is supposed to remove.

How to measure it without a single score

Score the conversation about the portfolio rather than the artifact itself: ask the same three questions (range, judgment, relevance) of every candidate, so the comparison stays structured even when the raw material isn’t. Where the role justifies the extra candidate time, pair it with a short take-home assignment closer to the actual work; that closes the comparability gap a portfolio leaves open.