01
The moment
A weekly meeting with the internal clinical team. Someone mentioned, in passing, how long the screenshots were taking.
Regulatory reporting required evidence: specific views from the platform, captured and assembled, on a schedule. The clinical team had been producing that evidence by hand — open a view, screenshot it, open the next one, screenshot that, and keep going until every required view had been captured. Then assemble the file and send it on.
Nobody had filed it as a problem. It was not on a roadmap, not in a backlog, not in a research finding. It was simply part of the job, and had been for long enough that it had stopped registering as a cost.
02
The recognition
Nobody in that meeting was describing a design problem. They were describing their week.
That is what makes a remark like this easy to miss. A complaint gets routed somewhere. An offhand comment about how something has always been done does not, because the person saying it is not asking for anything.
What I heard was a hard compliance requirement being met by hand, on a schedule, at a real cost in time and money. Manual, repetitive, rule-governed, and identical every time it ran.
That is not a workflow to improve. That is a workflow to remove.
03
The tool
I designed an administrative report tool built around a process runner. The clinician starts the report. The runner walks the required views, captures each one, and assembles the results. A zip file is ready to download seconds later.
Before — by hand
- Open the views one by one
- Capture them
- Assemble the file
- Send it on
Every step is the clinician's to do, on every report.
hours
After — one action
- Start the report
- The runner captures every view
- Download the zip
Nothing in the middle is the clinician's to do.
seconds
The design work was in the shape of it: one entry point instead of a procedure, and one artifact at the end that matches what the reporting requires. The steps in between stopped being anyone's job.
Anything that runs unattended against a compliance deadline also has to be legible while it runs. Minimize Surprise is at its most literal when the alternative is a person wondering whether a report they are accountable for actually finished.
04
Why it is here
hours → seconds per compliance report
This did not come out of a research session. There was no session. There was a meeting about something else, and a sentence that went by quickly.
User testing finds the problems you are looking for. Paying attention finds the ones nobody reports.
The methods for the first kind are well understood, and I use them. The second kind asks only that you keep listening after the agenda has moved on. It is where some of the best work is, because by definition nobody has asked for it yet.