UX clarity in MedTech vs FinTech: same word, different stakes

FinTech ships a one-tap send-money. MedTech ships a three-screen bolus confirm. Both call it "clarity." They mean opposite things and don't realize it. The word has no shared rubric across verticals, and that missing rubric costs real money in hiring rooms, RFPs, and design audits. This piece names a three-axis frame for fixing it.

The word "clarity" has no shared rubric

Picture a cross-vertical design review. Designer A demos a Tandem t:slim X2 bolus confirm: scroll the dose, see it re-displayed, tap to confirm. Three screens, deliberately. Designer B from FinTech: "Three screens for an 'are you sure?' Your activation funnel must be brutal." Designer A: "Your one-tap 'send $2,400 to @rikky_88' is malpractice. Where's the value re-display?"

Each is operating a different threat model. Each correctly applies their own rubric. The review ends with a polite "interesting, different context" and nothing changes.

I've sat in both reviews. The first time I got rejected at a MedTech portfolio screen, the feedback was "no IEC 62366-2 traceability — no Usability Engineering File output." I didn't know what those letters meant. Seven years of shipping FinTech onboarding, and the rubric I'd been measured against was suddenly the wrong one. The other direction is symmetric: MedTech designers apply to FinTech and get rejected for "no activation funnel metrics." Medical Technology Jobs lists IEC 62366 evidence as a portfolio requirement; Designity's FinTech hiring guide lists conversion metrics. Six to eight weeks of interview burn on both sides, and the actual gap — vertical-specific clarity rubric — never gets named in the feedback.

The rubric is invisible from inside. Each IC experiences their vertical's clarity as "what good UX is," not as a choice. That's the trap.

Three axes where "clarity" inverts

Three axes are enough to predict the rubric. Regulation is downstream — it falls out when you combine these three.

Stakes — worst-case error cost

What does the worst-case mistake cost?

MedTech default = morbidity or death. A wrong insulin bolus delivers irreversible drug into tissue within seconds. The reason Tandem and Omnipod engineer confirmation screens specifically to defeat tap-through habit is documented in insulin pump human-factors literature: confirmation appears only when a setting affects delivery, calibrated to break automaticity. The historical anchor is Therac-25 — a 1980s radiation therapy machine that killed patients because operator UI patterns let the system advance through inconsistent states. Modern MedTech UX engineering exists because of that file.

FinTech default = money or disputes. A wrong Zelle transfer triggers a dispute window: slow, painful, often unrecoverable, but not lethal. The confirm-modal is calibrated for Reg E disclosure compliance, not for breaking habit.

Except when the FinTech default is wrong. Alex Kearns, 20, took his life after Robinhood's UI showed him a -$730,000 buying-power figure with no surrounding context. The number was a margin-calculation artifact; he read it as actual debt. Robinhood changed the UI only after the wrongful-death suit. The Stakes axis had failed silently: the team had calibrated to FinTech's typical worst-case ($), not the actual worst-case for that user. The rubric still applies — it just has to be run honestly. If the worst-case is human, name it human.

Reversibility — time to irrecoverable

Compare two transfers. Wise lets you cancel before the recipient bank picks up — minutes to hours, well-documented. Zelle does not. The UI shape follows: Wise can lean on an undo-toast pattern; Zelle has to front-load disclosure because there is no undo to fall back on. CNN's August 2024 coverage of the $1B+ Zelle-scam losses names exactly that gap — users tap through habitually because nothing in the UI fights the habit.

MedTech's three-screen bolus confirm is the same logic in a tighter window. By the time the value has been re-displayed and the user has tapped a re-displayed number, the seconds-window is already gone. There is no equivalent of an undo-toast for delivered insulin.

So: how long until the user can no longer take the action back? That answer alone tells you whether your design has to fight habit or just disclose. LogRocket's reversible-actions framework and the academic UX Guardrails literature (transparency, proportionality, reversibility, contestability) both land on the same split.

Cadence — actions per user

Cadence compounds the first two. A high-cadence, high-stakes, low-reversibility action (bolus, 4–10× daily) is the worst possible automaticity trap, which is exactly why the confirm UI is engineered the way it is. A low-cadence FinTech transfer can carry a heavier disclosure load without retention damage. The same Reg E text that would gut a daily-use product is acceptable on a once-a-month wire.

What you do with this

What this gives you is not a portable design system. Last month I watched two senior ICs argue 30 minutes about a confirm modal. With this rubric we'd have spent five: "you're optimizing reversibility, he's optimizing cadence, here's the gap." That's the only win I'm claiming. Disagreements get faster, not resolved.

This won't make a FinTech designer safe to ship a bolus pump. It'll make her know she can't. Different problem, same value.

When NOT to apply: pure consumer products — social, gaming, content. Their stakes axis is engagement-defined, not harm-defined or money-defined. The rubric is built for products where mistakes have a defined cost in dollars, time, or harm. If your product's worst-case is "user closes the app," you are working with a different model.

Take it further

The full Clarity rubric: a 2×4 matrix mapping stakes, reversibility, cadence, and threat model to MedTech and FinTech defaults. Pasteable into Notion or Linear, droppable into Monday's design review.