Evaluate portfolios with a consistent, role-based lens

Choose what the portfolio can show

Decide whether you need evidence of problem framing, craft execution, system thinking, research, iteration, or stakeholder judgment. Do not ask one artifact to prove the entire scorecard. Tell candidates which capabilities will be reviewed and allow them to select work that demonstrates those capabilities, including anonymised or reconstructed material where appropriate.

Request context, not confidential detail

Ask for the problem, audience, constraints, team composition, personal role, and stage at which the candidate contributed. Make clear that private employer or customer information is not required. A candidate can explain reasoning with redacted visuals, process sketches, or a retrospective. Penalising responsible confidentiality selects for the wrong behaviour.

Separate work from presentation

A refined site can help navigation, but it may reflect time, coaching, or a specialist template. A simple document may contain excellent decisions. Review the chosen work before visual finish unless presentation itself is central to the role. If communication matters, assess whether the candidate can make the reasoning understandable, not whether the portfolio follows a fashionable style.

Use consistent prompts

Ask what changed after new evidence, which option was rejected, where the candidate disagreed with a partner, and what they would revise now. These questions reveal judgment and ownership. Record the artifact observation separately from the candidate’s explanation so the debrief does not blend assumptions with evidence.

• Problem and constraints understood
• Individual contribution isolated
• Decision rationale explored
• Reflection links to the work

Handle unequal portfolio access

Illustrative example: one product designer presents polished consumer launches; another cannot share client screens but provides a redacted service blueprint and explains three tradeoffs. The panel uses the same criteria and finds stronger systems reasoning in the second case. Visual availability does not become a proxy for ability because the process accepts alternative evidence.

Calibrate reviewers with real samples

Before live evaluation, have reviewers score one consented or fictional sample and compare evidence. Discuss where personal taste entered, what information was missing, and how ratings relate to the role. Recalibrate when the brief changes. Never circulate candidate work beyond the agreed hiring purpose, and avoid retaining materials longer than organisational practice permits. Decide how much time reviewers should spend and which pages or projects deserve attention. Unlimited browsing favours larger portfolios and makes comparisons inconsistent. If a required artifact is missing, request context through the agreed channel rather than searching for unrelated personal work. Record that request before rating.