Doctor, doctor!?

The research started with a research plan built around six questions from the brief, and then a service safari: I walked into a primary care centre (CAP) and followed the path a newcomer would, from the door to the first request for a health card.

What I found was a well-run centre. The information desk faces the entrance, the ticket keeps the queue orderly, and the receptionist is well prepared. The path fails at step five, and it fails the same way every time: to get a health card you need the padró, the municipal register of where you live, and to be on the padró you need an address someone will let you register.

The insight wall came from the workshop, desk research and the safari, and each note went through the same template: what I saw or heard, and why that was the case. Written that way, five separate observations turned out to have the same cause.

People are not waiting too long because they misjudge their symptoms. They wait because A&E is the only door that never asks for a card, so it becomes the door they use — late, and at the highest cost to the system.

I wrote three How Might We questions straight after the insights, and judged them on one thing: does the question accept the wall, or take it down?

Helping CatSalut identify vulnerable people fixes the institution's blind spot, but being identified is exactly what people fear. Reaching people without an address keeps the person at the centre, but works around the wall instead of removing it. The first question was the only one that went at the cause — and it held up, because the law already allows it.

Ideation ran as a Rapid 8 — eight ideas in eight minutes, volume over quality — and then onto a wall matrix: innovative or conventional, and whether each idea met our challenge or another one.

The matrix was humbling. Eleven of fifteen ideas fixed translation inside the doctor's office, a problem I had already ruled out in the insights. Only four went at getting into the system, and the strongest of those was not new at all: nonprofits holding an address for the people they support. It had been happening quietly in Barcelona for years.

The vision after research reads differently from the week-one version: a shared access service that opens the door to primary care before a crisis, for people in Barcelona who cannot register an address, through the organisations they already trust.

I explored six concept variations on a morphological chart — who holds the case, which channel, where the address lives — and dot-voted them down to three, which became one concept. It has three parts, one for each person in the service: a protocol for the nonprofit team, a case tool for the caseworker, and a message thread for Amina. Nothing for her to install, nothing to log into, and a clear promise about her data in the first message she receives.

The MVP is deliberately small: train one team, give them one number, run it for eight weeks.

This was the first time I designed something that was not a product, and the service blueprint is where that became concrete. Above the line of visibility there is almost nothing: one caseworker, one thread, in her language. Below it sit the Red Cross route for the card, the city's padró office, CatSalut, the CAP and the training that makes the nonprofit ready.

The blueprint also showed where the service would break: a CAP front desk that has never heard of the route, an address rejected by the padró office, and a padró wait that can run past six months. Those are the points the pilot has to watch first.

The boards above are rebuilt from the course canvases. These are four of the originals: the persona and journey map that put Amina and her caseworker side by side, the six concept variations and how they were scored, and the test kit for checking the riskiest assumption first — that people will answer a health message from a nonprofit they already know.