Hire designers for judgment, not portfolio polish

Hire designers for judgment, not portfolio polish

Hire designers for judgment, not portfolio polish

Design process

The short answer

A polished portfolio shows what a team shipped, not which calls the designer made. To see judgment, ask every candidate the same questions about their own decisions, add one short exercise that looks like the job, and score against a rubric that allows "not enough evidence."

A polished portfolio shows what a team shipped, not which calls the designer made. To see judgment, ask every candidate the same questions about their own decisions, add one short exercise that looks like the job, and score against a rubric that allows "not enough evidence."

Original mobile concept in charcoal and bright yellow-green. An app called Interviews shows a portfolio walkthrough kit: same questions for every candidate, project chosen by the candidate, 4 of 5 asked. Questions with follow-ups: What was your call? (What else did you consider?) What did you drop, and why? (Who pushed back?) Which constraint shaped it? (Was it real or assumed?) What would have changed it? (What did you test?) Which parts were yours? (And the team's?) The last one is not yet asked. Note: score each answer before the debrief; not enough evidence counts as a score. Buttons: Add note and Score answers. Beside the phone, a dithered exploded stack of three sheets: the top sheet has slid off to the side, leaving a dotted outline, and a solid yellow-green block sits exposed on the middle sheet. Text: Illustrative concept. Look past the polish. Ask every candidate about their own calls.
Original mobile concept in charcoal and bright yellow-green. An app called Interviews shows a portfolio walkthrough kit: same questions for every candidate, project chosen by the candidate, 4 of 5 asked. Questions with follow-ups: What was your call? (What else did you consider?) What did you drop, and why? (Who pushed back?) Which constraint shaped it? (Was it real or assumed?) What would have changed it? (What did you test?) Which parts were yours? (And the team's?) The last one is not yet asked. Note: score each answer before the debrief; not enough evidence counts as a score. Buttons: Add note and Score answers. Beside the phone, a dithered exploded stack of three sheets: the top sheet has slid off to the side, leaving a dotted outline, and a solid yellow-green block sits exposed on the middle sheet. Text: Illustrative concept. Look past the polish. Ask every candidate about their own calls.

Illustrative concept, not client work or measured results.

Polish isn't judgment

A clean case study is often a team's work, not proof of the designer's judgment. In a 2022 NN/g survey of 676 UX professionals, portfolios were the most cited assessment method and also "the most controversial," with some respondents struggling with "discerning the candidate's specific contributions or process on team projects."

Polish isn't judgment

A clean case study is often a team's work, not proof of the designer's judgment. In a 2022 NN/g survey of 676 UX professionals, portfolios were the most cited assessment method and also "the most controversial," with some respondents struggling with "discerning the candidate's specific contributions or process on team projects."

Ask the same questions about their own calls

Google's re:Work guide to structured interviewing uses "the same interview questions," an "identical scale," and a rubric for "poor, borderline, solid, and outstanding" answers.

Pick one project and ask every candidate: What was your call? What did you drop, and why? Which constraint shaped it? What evidence would have changed it? Which parts were yours? Follow up until you hear their reasoning, not the deck's. If an answer stays vague, write "not enough evidence" rather than a fail.

Add one exercise that looks like the job

NN/g says exercises "better demonstrate individual skills" and can be "fairly evaluated across all candidates." U.S. OPM guidance on work samples says they should mirror "the tasks employees perform on the job," and only test what people need on entry. Keep the exercise small, like a trade-off on a flow with a stated constraint, and score the reasoning, not whether they found your answer.

Ask the same questions about their own calls

Google's re:Work guide to structured interviewing uses "the same interview questions," an "identical scale," and a rubric for "poor, borderline, solid, and outstanding" answers.

Pick one project and ask every candidate: What was your call? What did you drop, and why? Which constraint shaped it? What evidence would have changed it? Which parts were yours? Follow up until you hear their reasoning, not the deck's. If an answer stays vague, write "not enough evidence" rather than a fail.

Add one exercise that looks like the job

NN/g says exercises "better demonstrate individual skills" and can be "fairly evaluated across all candidates." U.S. OPM guidance on work samples says they should mirror "the tasks employees perform on the job," and only test what people need on entry. Keep the exercise small, like a trade-off on a flow with a stated constraint, and score the reasoning, not whether they found your answer.

Illustrative concept: A judgment scorecard with example scores for one candidate. Five judgment signals scored as Poor, Borderline, Solid, Outstanding or Not enough evidence. Framing the problem: Solid. Trade-offs and what they dropped: Outstanding. Use of evidence: Borderline. Their call vs the team's: Not enough evidence. Knowing when to ask: Solid. Note: same rubric for everyone; score before the debrief; not enough evidence is a score, not a fail.
Illustrative concept: A judgment scorecard with example scores for one candidate. Five judgment signals scored as Poor, Borderline, Solid, Outstanding or Not enough evidence. Framing the problem: Solid. Trade-offs and what they dropped: Outstanding. Use of evidence: Borderline. Their call vs the team's: Not enough evidence. Knowing when to ask: Solid. Note: same rubric for everyone; score before the debrief; not enough evidence is a score, not a fail.
Illustrative concept: A judgment scorecard with example scores for one candidate. Five judgment signals scored as Poor, Borderline, Solid, Outstanding or Not enough evidence. Framing the problem: Solid. Trade-offs and what they dropped: Outstanding. Use of evidence: Borderline. Their call vs the team's: Not enough evidence. Knowing when to ask: Solid. Note: same rubric for everyone; score before the debrief; not enough evidence is a score, not a fail.
Illustrative concept: A judgment scorecard with example scores for one candidate. Five judgment signals scored as Poor, Borderline, Solid, Outstanding or Not enough evidence. Framing the problem: Solid. Trade-offs and what they dropped: Outstanding. Use of evidence: Borderline. Their call vs the team's: Not enough evidence. Knowing when to ask: Solid. Note: same rubric for everyone; score before the debrief; not enough evidence is a score, not a fail.
Illustrative concept: A judgment scorecard with example scores for one candidate. Five judgment signals scored as Poor, Borderline, Solid, Outstanding or Not enough evidence. Framing the problem: Solid. Trade-offs and what they dropped: Outstanding. Use of evidence: Borderline. Their call vs the team's: Not enough evidence. Knowing when to ask: Solid. Note: same rubric for everyone; score before the debrief; not enough evidence is a score, not a fail.

Before your next design interview

  • Write down the judgment signals the role needs.
  • Ask every candidate the same questions, and separate their decisions from the team's.
  • Use one short exercise that mirrors the work. Test only what they need on day one.
  • Score before the debrief. "Not enough evidence" is a real score.

Before your next design interview

  • Write down the judgment signals the role needs.
  • Ask every candidate the same questions, and separate their decisions from the team's.
  • Use one short exercise that mirrors the work. Test only what they need on day one.
  • Score before the debrief. "Not enough evidence" is a real score.