Back to Insights
July 16, 20265 min readInsightconstruction softwareDesignHealthcaremarketing

Watching or Working

Until this week, the console that files all my Medicare claims had no way to sign out. Plain forms, no animation, not one hour of visual design. Meanwhile the marketing site for the same product got choreographed scroll animations and hand-built React recreations of those very screens. The copies are prettier than the originals. That sounds like a scandal. I think the money went exactly where it should have, and most software points it the other way.

01Where the polish went

clearpathcare.ai is the prettiest thing I have ever shipped. Every scroll position is choreographed with GSAP and Lenis. Every console screen was rebuilt by hand in React so it could slide in at just the right moment. Days of art direction for a page most people will see once.

The console behind it, the tool that enrolls patients, logs the calls, and files the claims, got none of that. Zero hours of visual design. Not because it matters less. It matters more; it is the part that makes the money.

The difference is what the person in front of each screen is doing. A visitor to the site is leaning back. They are judging, and they decide in about thirty seconds whether you are real. Looking good is the whole job. A biller in the console is leaning in. They are there all day, and every animation between them and the claim is friction. The best thing that screen can do is get out of the way.

02The koi fish

I know what happens when you get this wrong, because I got it wrong once. MOTOR is a quoting engine I built for construction contractors. At some point I gave the quoting flow a koi companion: a little animated fish that swam along while you built a quote. I thought it was charming.

Contractors killed it inside three quotes. The feedback was not a design critique. It was "pretty annoying" and "just give me the quote."

What broke

MOTOR's quoting flow got an animated koi fish to keep users company. Contractors killed it inside three quotes. The verdict was "pretty annoying" and "just give me the quote." Decoration inside a work tool is a tax you charge on every single use.

Just give me the quote. That sentence taught me more about design than any book. A worker does not experience your polish as a gift. They experience it as one more thing standing between them and being done.

03The rule

So here is the rule I use now. Do not ask how important the system is. Ask what the person in front of it is doing. Watching? Spend on looks. Working? Spend on speed.

By that rule, the console being the most important thing I own and the plainest is not a contradiction. It is the rule working.

Key insight

Spend design where people watch and speed where people work. The budget follows what the user is doing, not how much the system matters.

Polish is for people who are watching. Speed is for people who are working.

This does not mean work tools get no design. This week the console finally got real design attention, and every bit of it was invisible: the browser back button works now, breadcrumbs show where you are, a confusing device field got a sane default and a plain hint. Nobody will screenshot any of it. Everyone who works in it will feel it.

And when the console did need to be watched, because prospects wanted to see it run, I did not decorate the console. I bolted a guided demo onto the side: nine chapters, about forty steps, strictly read only. The show got its own stage. The tool stayed plain.

The result

The console's guided demo shipped as nine chapters and roughly forty read-only steps. The show got its own surface, and the screens people actually work in stayed plain and fast.

3
quotes before contractors killed the koi fish
0
hours of visual design in the console that files the claims
9
chapters in the guided demo, a surface built for watching
40
steps of polish kept out of the working screens

04The test

I run systems in healthcare, in construction software, and in marketing, and the same one-line question now gates every design hour in all three: is the person on the other side of this screen watching or working?

  1. 1Ask watching or working before you spend a single design hour.
  2. 2Spend looks on screens people judge and leave. Spend speed on screens people live in.
  3. 3Work tools still deserve design, the invisible kind: a back button that works, a sane default, a breadcrumb that tells you where you are.
  4. 4If a work tool needs to be watched, build the show a separate surface, like a demo, and keep it out of the flow.
  5. 5When a working user calls something pretty annoying, believe the annoying, not the pretty.

Step back far enough and the pattern is simple: the closer a screen sits to the money, the plainer it gets. If the part of your product that does the real work looks embarrassingly plain, you are probably spending right. And if you run an ugly system that quietly earns, I would like to compare notes, because we are making the same bet. The money is made in plain rooms. The pretty rooms are how people find the door.