Every Pixel Accounted For

There is a strange inversion at the center of modern technology. The applications we use to order coffee, hail a ride, or scroll through photographs are refined to an almost obsessive degree. Every animation timed, every margin measured, every word on every button argued over. The software a physician uses to decide what medication to give you, at what dose, at two in the morning at the end of a twelve-hour shift, frequently is not. The stakes could not be more different. The care taken has been, for decades, exactly backwards.

In a 2019 Mayo Clinic study, physicians scored their electronic health records a 45.9 out of 100 on the System Usability Scale; the same standardized measure used across the software industry. That is a failing grade. It places the average medical record in the bottom nine percent across more than 1,300 prior studies using that scale, below the everyday consumer software most people never think twice about. The tools entrusted with the highest-stakes decisions in our lives test worse than the tools we use to split a dinner bill.

This is not a cosmetic complaint. It is a clinical one.

The stakes are inverted

In consumer technology, the cost of friction is small and forgiving. A confusing checkout loses a sale; a cluttered screen loses a minute; a frustrated user closes the tab and tries again tomorrow. Designers work relentlessly to eliminate that friction anyway, because attention and delight are the only things they compete on.

In medicine, the same friction is measured in a different unit. Researchers have counted roughly four thousand mouse clicks in a single emergency-room shift. One widely cited study found physicians spending 5.9 hours of an 11.4-hour workday inside the electronic record and 5.1 hours with actual patients. When a screen is cluttered, an allergy warning can be missed. When a field is ambiguous, a decimal can move. Here, a design decision is not a preference. It is a clinical decision, made once by a designer and then re-lived thousands of times by the people who use it.

Why the most important software became the least considered

It is worth asking honestly how this happened, because the answer shapes the cure. Healthcare software did not end up this way because the people who build it are careless. It ended up this way because of who it is sold to.

Most enterprise medical software is bought by committees and procurement departments, not by the clinicians who will spend their working lives inside it. It is evaluated on feature checklists, compliance requirements, and contract terms, on what it can be said to do, not on how it feels to do it. Once installed, it is protected by switching costs so high that dissatisfaction rarely becomes departure. The competitive pressure that forces a consumer app to be genuinely good (the pressure of a user who can leave in a single tap) has simply been absent. Software gets better when its makers have to earn the user’s attention. For a long time, in this industry, they have not had to.

Every Pixel Accounted For

It is fashionable to say that companies like Apple and Google are good at design because they make things beautiful. That misses the point. What they understand is that the interface is not a layer applied on top of the product. It is the product. The entire surface of the relationship between a person and what the software can do for them. So every pixel is accounted for, because every pixel is a decision about where a human’s limited attention will go.

There is research beneath the instinct. As early as 1995, studies of interface design found that people rate attractive systems as easier to use and that the correlation between beauty and perceived usability was stronger than the correlation between beauty and measured usability.

We extend trust to things that appear considered.

In a consumer app, that trust sells a subscription. In a clinical tool, that trust is the difference between a clinician who leans on the software and one who quietly works around it.

Medicine does not deserve this discipline less than a music app does. It deserves it far more.

This is the conviction we design from: in healthcare, care for detail is care for people, and the two cannot be pried apart. The wording of a label, the alignment of a number, the contrast on a warning, the order in which fields are asked for. These are the things that decide whether a tired human being catches the one detail that matters at the end of a long night.

Beauty, in this context, is not ornament. It is reduced cognitive load. It is the quiet confidence that lets someone think about the patient instead of the machine.

Restraint is as much a part of this as refinement. Much of what has gone wrong in clinical software is the result of adding more alerts, more fields, and more mandatory clicks until the signal drowns. Good design is not the art of putting everything on the screen. It is the discipline of deciding what deserves a person’s attention, and having the restraint to leave the rest off.

Design is how we give clinicians their time back

There is a human cost to all of this that the usability scores only hint at. Documentation burden is now among the most frequently cited drivers of physician burnout. More than a fifth of physicians report spending over eight hours a week on the record after hours, the charting that follows them home from the hospital. Every unnecessary click is a small tax on a person’s evening, and on the attention they have left for the next patient.

The encouraging news is that this is solvable, and the evidence is beginning to arrive. In a 2025 quality-improvement study across six health systems, burnout among participating ambulatory clinicians was 51.9 percent before ambient AI scribe use and 38.8 percent after 30 days. The before-and-after shift suggests that easing documentation burden may be felt quickly. That is the whole case for taking design seriously here: not that it is pleasant, but that it hands time, focus, and attention back to the people we rely on to keep us well.

Our philosophy

So yes, we strive to account for every pixel. We argue about spacing and wording and the exact colour of a warning as though they matter, because in this domain they do. We treat the interface as an instrument, the way a surgeon treats the balance of a scalpel, because the person on the other side of the screen is making decisions far more consequential than any we make in designing it.

The goal is not software that impresses. It is software that disappears, that asks for nothing it does not need, surfaces exactly what matters, and lets a clinician spend their attention where it belongs: on the patient in front of them. Healthcare has tolerated its most important tools being its least considered for a very long time. We think that is not only a design failure but a moral one, and it is the one we strive to correct. Every pixel must be accounted for. Every pixel is a clinical decision.

We are not there yet

We owe you one more thing, because a philosophy is worth little without it: an honest account of where we actually stand. What we make today is, we believe, meaningfully better than what it aims to replace. But the incumbents were never our benchmark. This vision is, and measured against it, we still fall short. We would rather say that plainly than pretend otherwise.

We will admit something less comfortable, too. We do not yet know exactly what the finished version of this looks like. We can see its shape; the calm, the clarity, a tool that disappears in the hand; but some pivotal piece, most likely a technological one, still feels missing, just past the edge of what anyone can build today, even despite the AI revolution. So we keep moving toward it: refining, discarding, rebuilding, chasing a vision we can picture clearly and a final form we can still only sense.

That is the honest promise underneath all the rest, not that we have arrived, but that we will not stop walking toward it.

We are not there yet. We intend to be.

References

1. EHR usability scored a failing grade (mean SUS 45.9). Melnick et al., Mayo Clinic Proceedings (2019) · Healthcare IT News summary
2. ≈4,000 clicks per ER shift; 5.9 vs. 5.1 hours of an 11.4-hour day. “Death by 1,000 Clicks,” KFF Health News / Fortune (2019)
3. After-hours “pajama time” documentation (2024 physician data). American Medical Association
4. Alert fatigue and high override rates. AHRQ PSNet — Alert Fatigue primer
5. Aesthetic-usability effect (Kurosu & Kashimura, 1995). Nielsen Norman Group
6. Burnout was 51.9% before and 38.8% after 30 days of ambient AI scribe use. JAMA Network Open (2025)