A clinical product is read in a waiting room. The light is bad, the time is short, and the stakes are not abstract. We spent six weeks shadowing patients and clinicians before we drew anything.
Most of what we saw was not a design problem. It was an inventory problem: too many fields asking for the same thing twice, too many screens between a patient and the one decision they came to make.
The rule we wrote
If a patient could not finish a task without asking for help, the screen was wrong, not the patient. We carried that rule into every review.
The measure of success was whether a patient, sitting in a waiting room, could finish a task without asking for help.
We rebuilt the product around five tasks. Everything else fell away. The design followed the tasks, not the other way around.
What we would do differently
- Start the shadowing earlier, before the brief is signed.
- Write the tasks on a wall before opening the design file.
- Measure task completion, not page views, in the first usability round.