Skip to content

Slide UX · agency engagement

American Board of Radiology

Replacing a ten-year in-person exam with something radiologists could do in a few minutes a week.

Role
UX Designer two-person design team
Client team
Director of Analysis Director of IT, developers
Scope
User flows, wireframes, visual design, prototype, usability testing
Duration
About one year initial engagement

A note on credit. This was agency work at Slide UX, delivered by a two-person design team with a senior designer leading and me as the second designer, alongside ABR’s own analysis, IT, and development staff. The decisions below were the team’s. I’ve written up the parts I worked on and what I took from them. Client artifacts remain the client’s, so the visuals here are representative rather than production screens.

01The problem

Certification that cost a decade of dread

The American Board of Radiology certifies physicians in radiology. Maintaining that certification meant an in-person exam once every ten years, which radiologists had to travel for and study heavily to pass.

The complaint was not that the exam was hard. It was that it tested the wrong thing. An academically weighted test taken once a decade is a poor instrument for measuring whether a physician is practicing well today, and it rarely touched the working knowledge they actually used.

The shift

One exam every ten years became fifty-two questions a year

Roughly one short question a week, taken wherever the radiologist already was. Same purpose, an entirely different relationship to the physician’s time.
02Architecture

A simple surface over a lot of moving parts

A quiz looks trivial. A question is asked, an answer is chosen, a result appears. That is exactly how it should feel, and almost none of the work is there.

Underneath it we had to settle how answers were tracked over a year, what happened when someone got one wrong, which subject bank a question was drawn from, how imaging and other media were handled inside a question, and what the experience did when someone logged in on a device the tool could not support.

Sequencing

Every path mapped before a single wireframe

We worked through the full user flow with ABR first. Deciding those branches on a flow diagram is cheap. Discovering them halfway through visual design is not.
03Data visualization

Progress you can read in a glance or not at all

The hardest single component was the graphic showing a physician where they stood for the year. It had to carry four things at once: how much they had completed, the trajectory that implied, whether they were at risk of missing the requirement, and whether they had already failed for the year.

The ABR palette was almost entirely shades of blue, which is a real constraint when the whole job of the graphic is separating four states fast. Blue-on-blue differences that look fine at full attention disappear at a glance, which is the only way this chart would ever be read.

The fix

Banded zones and a line that changes color

At-risk and danger-of-failing zones were drawn into the chart itself, and the tracking line shifted from blue to red as it crossed into them. Status became a property of the picture rather than something to work out from it.
04Validation

Six radiologists, one thing nobody predicted

We ran a usability study of six one-hour sessions with board-certified radiologists, against a prototype of the initial designs.

Most of it confirmed the direction. Participants were genuinely enthusiastic about the premise, relieved at not having to travel for a heavy academic exam, and found the flow intuitive. The progress data landed.

They didn’t want to look at a white screen. Radiologists spend most of the day in a dimly lit room.
The finding that changed the design

What it changed

The whole thing went dark

Not a preference toggle bolted on at the end. The designs were reworked around the lighting conditions these users actually work in. It is the clearest example I have of something no amount of desk research would have surfaced, and six conversations did immediately.
05Craft

No training, because there was no room for any

Radiologists needed to study the imagery in a question the way they would study a real case, which meant the tools for doing that had to be present in the interface. The risk is obvious: every control you add is attention spent on the tool instead of the question.

The tooling was placed close to the asset it acted on, and used icons already familiar from other ABR products, so recognition did the work that a tutorial otherwise would.

06Outcome

The client came back

The quiz replaced a decade-long cycle of preparation and travel with something that fit into a working week, and it measured applied knowledge rather than recall.

ABR returned to the team for further work: a remote oral exam, the certification exam, key pages of their intranet and volunteer hub, and additional usability studies.

The dark mode finding is the one I still bring up. The design was good, the flow tested well, and the thing that mattered most was a detail about the room the work happens in.

Next project

Clarix→
All work