How to run the persona-deck workshop with undergraduates in 3.5 hours — and what surprised us across ten cities.
The brief sounds modest: give each student a card, let them interview the person on it, and listen. In practice the workshop does something more stubborn. It teaches a cohort, in a single afternoon, that empathy is a skill you can get better at by narrowing your attention — and that the narrowing is what makes the listening possible.
We call it Empathy with a Lens because the SDG-informed addendum to the interview adds a second focal length: the student first meets the persona as a person, and only then is asked to see them through a specific lens — access to clean water, decent work, quality education, life on land. The two focal lengths argue with each other. That argument is where the learning sits.
Three and a half hours is enough for a single pass through the full cycle. We budget the time roughly like this:
The timings are deliberately tight. Students who have run other design-thinking workshops arrive expecting slack. Removing the slack is part of the pedagogy — the constraint forces them to notice what they are actually doing with their attention.
The handouts are not scaffolding in the usual classroom sense; they are the workshop's instruments. H3 forces students to commit to ten observations before they are allowed to ask the AI a single question. H4's nine-branch mindmap slows the interview down on purpose: each new question has to plant itself on a named branch, so students cannot drift into the kind of generic "tell me more about your day" rambles that feel empathetic and reveal nothing. H5 imposes a synthesis discipline that individuals, pairs, and quartets each have to pass through without skipping a step.
The facilitator's job is to guard the handouts. Not to add content, not to elaborate, not to rescue a struggling group. The scaffolding is the intervention.
The cards are the smaller half of the work. We expected the persona deck to be the centre of gravity. It turned out to be the least contested element. What students argued about, week after week, was the ten-dimension profile they had to draft before interviewing. The forced pre-commitment changed the interview from a discovery exercise into a hypothesis-testing one, and that shift — not the card — is what most students named when we asked them later what had changed.
The AI is a better listener than the student, at first. In the first hour, students ask the AI questions that are looser and more generous than they would ask a classmate playing the same persona. Something about the non-judgmental surface of an AI model opens the interview up. By the third hour the asymmetry has reversed — the students are asking sharper questions than the AI can productively answer, and the exercise moves into synthesis. This arc repeated in every cohort we ran.
Cross-city cohorts find different edges. Run in Helsinki, students gravitated toward life on land and decent work; run in Kuala Lumpur, quality education and reduced inequalities dominated; in Townsville, clean water and climate action surfaced almost automatically. The cards are the same in every city. The lenses students pick up are local, and that is fine — the workshop was never meant to produce a single answer.
The failure mode is always the same. Cohorts that skip H3 (the ten-dimension pre-commit) produce charming interviews and thin briefs. Cohorts that do H3 but skip the individual-first step of H5 produce clean briefs but shallow consensus. In both cases the handout they skipped was the one they told us felt most "optional" at the time. It never is.
Print more cards than you need. Read H3 out loud, don't just hand it out. Refuse to explain the mindmap in H4 — let students discover it. Hold the quartet debrief to fifteen minutes even if groups are still talking; the truncation is generative. And — the detail that matters most — keep a quiet note of which SDG lenses the cohort gravitated toward. Over a semester the pattern will tell you something the students cannot tell you themselves.
Hansen & Kaskenpalo · March 2026