Notes · Point of view · 8 min read

Nobody wants a lab report

I once spent a year turning 115+ biomarkers into a product, back when I designed for health. The main thing I learned: people don’t open a health app for data. They open it with a question, and most apps answer with a spreadsheet.

01 · The filing cabinet

Apple Health is complete, accurate and unread

Apple Health might be the most impressive database ever shipped to a phone. Every metric, every source, every unit, neatly filed. And it behaves exactly like a filing cabinet: you go there when you need a document, then you leave. Steps, distance, a bar chart, a favourites list you curated once and forgot. It presents data with total fidelity and answers no question a worried person actually has.

That’s not a swipe at the designers; it’s a platform, not a coach, and it’s honest about that. The problem is how many health products copy its shape while claiming to be a coach. A grid of metric tiles is not an answer. It’s homework.

Open the Edit Favorites screen and the filing-cabinet nature becomes explicit: Active Energy, Cycling Distance, Downhill Snow Sports Distance, NikeFuel, Pushes, Stand Minutes, a taxonomy with a star next to each entry. It’s a beautifully organised warehouse, and asking a worried person to curate their own health dashboard from it is asking a patient to assemble their own chart. Strava has a milder version of the same disease: the Progress tab is charts of training load over time, faithfully plotted, with no sentence underneath telling you whether the line means you should rest or push.

None of this happens because designers love spreadsheets. It happens because a verdict is expensive and a number is cheap. Shipping a chart is safe: the data said it, not you. Shipping a sentence like “you’re fine” means someone in the building took responsibility for being wrong, ran the wording past legal, argued with a clinician and won. A chart hedges; a sentence commits. Most teams, given the choice, hedge, and then wonder why nobody opens the app twice.

02 · The translators

Whoop and Oura sell interpretation, not numbers

Whoop’s home screen leads with derived scores: 58% recovered, 4.9 strain, 40 HRV. That’s a real improvement over raw data, and it’s also a bargain with a catch: you traded one literacy problem for another. Now you’re learning Whoop’s private language, and until you do, the numbers carry authority without meaning. A percentage feels precise even when you have no idea what it’s a percentage of.

Oura is gentler with the same material. Scores come wrapped in plain words, and even its empty state works: “Waiting for your sleep data” keeps the promise visible instead of showing a zero. The difference sounds small. To an anxious person at 7am, a zero and a “waiting” are entirely different messages.

Apple Health iOS summary screen

Apple Health: perfect filing, no verdict

Whoop iOS home screen with recovery, strain and HRV scores

Whoop: scores with authority, in a private language

Oura iOS home screen

Oura: plain words, and an empty state that keeps the promise

Reference screens via Mobbin

People don’t open a health app for data. They open it to hear “you’re fine”, with evidence.

The lab report on one side, three answers in order on the other: that is the whole job
03 · When words do the work

The best health screen I know is two words long

Oura’s readiness card, at its best, reads: 68. Fair. Go easy. Then one plain sentence about resting heart rate, and done. That’s the whole answer stack in one card: the score is the evidence, “fair” is the translation, “go easy” is the instruction. Nobody needs to know what HRV stands for to act on it correctly. Oura has published how the readiness score works for people who want the workings, which is exactly where the workings belong: one tap away, not on the front page.

Compare the redesign story on the other wrist. When Whoop rebuilt its home screen around three dials, the company presented it as radical simplification, and reviewers still asked whether it was actually intuitive. Three dials is better than thirty tiles, but a dial is still a number wearing a costume. The unit of meaning in a health product is not a gauge. It’s a sentence a tired person can obey.

Building Emerald taught me to write for the 2am reader specifically. The medical team wanted precision, rightly, but clinical precision reads as menace when you’re scared: “abnormal” is a technical term on one side of the screen and a siren on the other. We’d rewrite until the words did the same clinical work without the adrenaline: slightly elevated, common, worth mentioning at your next check-up. Same fact, different heart rate. It took longer than designing the charts did, and it mattered more.

This is also where my ads instinct, the one from every screen is an ad, needs a leash. In performance creative you amplify emotion; in health you damp it. Same two-second glance, opposite goal: the glance should end in calm, not a click.

Oura readiness card reading 68, Fair, Go easy

Oura: score, translation, instruction. The whole answer stack in one card

Apple Health edit favorites list of dozens of metric types

Apple Health’s taxonomy: curate your own chart, Downhill Snow Sports Distance included

Strava progress tab with training load charts

Strava: faithful charts, no sentence telling you whether to rest or push

Reference screens via Mobbin
04 · The order of answers

Three questions, in this order

Building Emerald, we compressed 115+ biomarkers into one score with full reports underneath, and the structure that survived every round of testing was a hierarchy of three questions. Am I OK? What changed? What should I do next? Answer those, in that order, above the fold. Everything else, the charts, the reference ranges, the sources for the sceptics, is the appendix, and it should be one tap away rather than on the front page.

Health data also gets read differently than any other data. A confusing bank chart makes people annoyed; a confusing health chart makes them frightened, at scale, at 2am. Calm beats clever every time. If a screen can’t say “you’re fine” or “here’s the one thing to do”, it should say nothing yet, plainly, the way Oura’s waiting state does.

The same hierarchy discipline applies wherever people arrive with a question and a pulse above baseline: money works this way too, which is why the argument in what most fintech onboarding gets wrong is really this argument wearing a suit.

Further reading, if this is your problem space:

More notes