Putting the plan before the login wall

A journey-mapping study had already named first-time mobile entry the deepest drop-off of any patient type: two screens of account setup, one literally branded "Welcome to Fullscript's Dispensary," before a new patient ever saw the plan their provider built for them. Patients told researchers they saw the whole product as a one-time fulfillment site.

For the first time in a long while, product and design got to define the scope before engineering did, and I didn't want to waste it. Four days in, I ran a 90-minute workshop — what does great onboarding do for patients, practitioners, and the business; which entry point is V1; dot-vote; go.

A patient arriving with an unpurchased plan needs something different from one who already ordered, or one mid-way through an intake. The flow reads the patient's state and routes accordingly — the unpurchased-plan patient goes straight to their plan, everyone converges on the same health-profile spine after.

The team's own playbook says never trap a patient in a guided flow without an exit. But the goal question the flow was built around depends on a genuine answer, and an escape hatch invites the "I'll do this later" non-answer that defeats the point. I removed skip on the goals screen alone — the sanctioned out is a "Prefer not to say" option — and kept every other screen skippable.

Goals, age and sex, wearable connection, and routine setup could have shipped as one "complete your profile" step. I gave each its own screen instead, one question at a time. That meant more screens to build, test, and instrument, and the whole flow still finished at a 1.1-minute median, well under the 2-minute target.

The old flow asked for notification permission the instant a patient logged in, and converted 13.5% of asks. The product carried a scar here: an earlier feature had died not because practitioners rejected it but because an unrequested email fired on every edit — noise killed the loop. The lesson stuck: notification design is the product.

A phased rollout would have been easier; our analytics partner made the case for a true holdout instead, and he was right. The flow went live in June to a quarter of patients, with a matched control who never saw it — the only reason the results below are trustworthy.

Two misses: notification enablement regressed — 9.8 per 100 against the old flow's 13.5 — because 92% of patients never scroll to the buried ask, though 97% of those who reach it say yes. The platform's numbers make that costly: a plan notification is the best-performing message it sends, clicked about fifty times more often than a promotion. And only a third of unpurchased-plan patients finish the flow — for non-converters it ends where the purchase decision does.

The 40-day reorder analysis came back clean, and at the end of July the team aligned on releasing the flow to every patient. The open questions have become other teams' roadmaps: a re-entry path for patients who abandon at the purchase decision, merchandising against the stated-goals data, and where the notification ask earns its best yes.

Web still has no equivalent entry experience, and the platform has since sized that gap: patients who start in the app reorder about 12% more often than web starters, and they place more orders a year rather than bigger ones. The difference is frequency, not basket size — which is exactly what an entry experience moves.

Getting patients to their plan fast was the right bet, and for the two-thirds who don't buy, it's also where the flow ends. Next time I'd design the way back in at the same time as the way through.