Habits take months. Every other tool stops in week one.

The evidence for building behaviour over months instead of blasting notifications for a week, and the model you can argue with.

Every retention tool sells a slower leak.
We sell the floor.

Retention curves always fall. What separates a business from a corpse is whether the curve flattens, and the thing that flattens it is habit.

Months, not a week.

Habits build fast at first, then level off. Across the research, it takes two to five months before showing up stops being a decision and becomes something people just do, and how long varies enormously from person to person.

Past that line, coming back isn't a choice they make anymore. It's just what they do. That's your plateau. That's product-market fit, made of people.

Every nudge tool on the market fires for a week and stops. A boost that ends in week one cannot move a floor that takes months to build. Worse: hammering people with reminders to pump up week-one numbers causes the fatigue and uninstalls that wreck the very floor you were trying to raise.

Singh et al., 2024, systematic review and meta-analysis, 20 studies, 2,601 participants: habits consolidate over two to five months, accruing gradually rather than at a threshold. Lally et al., 2010 gives the curve shape and a 66-day median, and found that missing a single day did not measurably harm habit formation, which is why our streaks forgive you. Because time-to-habit varies so widely, we tune the taper per user rather than treating 66 days as a rule.

~day 66 median. varies widely automaticity time install everyone else stops here

The floor is reachable. Real apps already hold it.

The best health apps hold 40%+ at day 30, specifically the ones building genuine behaviour change rather than engagement tricks. Mental-health apps with real touchpoints hold 20–35% at day 90, while pure tracking apps with no touchpoints sit below 10%. Duolingo cut monthly churn from 47% to 28% on streak-and-reminder loops.

None of that is us, it is what the field has already proven possible. The gap between a sub-10% floor and a 30% floor is not a messaging problem. It is a behaviour problem, and it is the one thing we do.

2–3×

Catch the drift, not the corpse

Automated triggers fired at early dormancy produce 2–3× higher return rates than waiting a week. The recovery window closes in 3–7 days.

  • Six signals, scored against each user's own rhythm
  • A weekly user isn't drifting on day 2, we know the difference
  • Fires before they're gone, not at the funeral
15–30%

Mechanics encoded, not configured

Streaks, progress and loss-aversion mechanics raise day-30 retention by 15–30%. There's no dashboard to master and no campaign to build.

  • Loss aversion beats reward, every time
  • Forgiving streaks: one missed day doesn't break a habit
  • The taper: we go quiet once the habit holds, tuned per user, not a fixed date. Nobody else stops at all.

Where does your curve flatten?

Drag your numbers in. This is our behavioural model run on published benchmarks, not our results, and not a promise. The day you install, it is replaced by a prediction built from real apps shaped like yours.

Industry medians: D1 ~25%, D7 ~8%. If you don't know yours, that's the first thing the free tier tells you.
Doing nothing — predicted plateau
—
With the 66-day arc — modeled
—

Run it against your own curve

The model on this page is built from published rates. Your dashboard replaces it with your own numbers in the first week.

Request access