Consumer appHealth and fitnessWin-backBuilt on CleverTap
Same lapsed users. Same channels. Revenue up 79%.
Healthify's ex-premium users had paid for coaching, seen progress, then let their plans lapse. The comeback messages they got could have gone to anyone. This is the engine that made each one theirs.
The product
Healthify's paid coaching plans for weight loss and medical conditions.
The audience
Ex-premium users. People who paid, saw results, and let the plan lapse.
The gap
Every lapsed user got the same message. Nothing in it was about them.
The bet
Their own progress would bring them back faster than any discount.
Pain point
These weren't cold users. They'd already paid once, and most had real progress to show for it. But the reactivation messages treated them like strangers: one template, one offer, sent to the whole segment.
The data to do better already existed. Every user had months of logged weight, workouts, meals and app activity. None of it reached the message.
The reach was there. The messages just had nothing personal in them. A lapsed user has no reason to open a generic nudge, and every ignored message makes the next one easier to ignore.
The strongest reason to come back was already sitting in their history.
What I did
- 01
Picked the signals that mattered.
Weight change, workout days, calories burned, food logs, app launches and plan length. Six data points that tell a user's story better than a segment name does.
- 02
Built logic that turns data into sentences.
A formula layer mapped each user's raw numbers into readable, personal lines, so every message carried that person's own progress.
- 03
Designed for missing data.
Not every user logs everything. When a signal was blank, the message fell back to the next strongest one instead of breaking.
- 04
Proved it, then scaled it.
It launched in India first. Once the numbers held, the same engine rolled out to NRI users.
Before and after
| Before | After | |
|---|---|---|
| Message content | One template per segment | Built from each user's own progress |
| Data used | Segment label | Six behavioural signals per user |
| Missing data | Not handled | Automatic fallback to the next signal |
| Markets | India | India and NRI |
Results
Click-through rate, 0.65% to 2.20% by the NRI rollout (1.37% in India)
Leads, 245 to 544
Revenue, ₹3L to ₹5.37L on the India cohort
Return on ad spend, 13 to 20 by the NRI rollout (18 in India)