./pulse

a health tracker that cares about day 30, not day 1.

role:

senior pd · 0→1

team:

1 pd · 4 eng · 1 pm · 1 coach

timeline:

18 months → jan 2024

platform:

ios + android

[ pulse — 3 app screens on device frames ]

$ cat ./context

the problem

71% of users were gone within a week.

northcare’s pilot habit app had a brutal day-7 cliff. the pattern was always the same: a user missed one day, the streak reset to zero, and they never came back.

habit apps are built to punish lapses — but lapses are how habits actually form. the product worked against its own users.

the brief >

“build the first health tracker that forgives you.”

2.4×

day-30 retention vs. pilot app

61%

weekly habit completion across all cohorts

4.8★

app store rating — 12k ratings, year one

1/day

notification cap — a feature, not a limit

$ cat ./process

research → flows → iterations

(a) research

lapses didn’t cause churn. highlighting them did.

a two-week diary study with 21 users, plus working sessions with a behavioural coach. users who lapsed but weren’t shown the lapse recovered at nearly normal rates. the red broken-streak screen was the churn moment.

[ diary study synthesis — lapse timeline ]

[ flexible streak model — flow diagram ]

(b) flows

weekly targets instead of daily chains.

“flexible streaks”: hit your target 5 of 7 days and the streak holds. one habit on screen at a time. progress framed as accumulation, never as loss.

three beta cohorts. guilt-red died in beta 2.

200-user cohorts with weekly retention readouts. the alarm-red “missed” state tested worse than showing nothing at all — we replaced every punitive state with a neutral one and watched d7 climb.

[ beta 1 vs beta 2 — missed-day states ]

$ open ./solution

what shipped

[ today screen — final ui, one habit at a time ]

[ flexible streak system ]

streaks that bend instead of break — 5-of-7 targets, weekly resets, zero red.

[ notification strategy map ]

one nudge per day, timed to each user’s past completions. opt-out kept front and centre.

$ tail ./outcome

after launch

pulse launched jan 2024 and became northcare’s most-rated product in nine months. the flexible-streak model was later adopted across their whole app family.

“the first tracker i didn’t delete in january.”

— beta user #117, diary study exit interview

// next case — 03/04

./relay →

back to index ↑