From Leaking Retention to 58% D30: A Digital Banking Turnaround
A digital banking app with a live MoEngage instance was seeing D30 retention below 33% and rising push opt-out rates. We rebuilt their lifecycle from the ground up: new event taxonomy, new segmentation, new journey architecture. Then we shipped 12 journeys in 90 days.
By the numbers
What we walked into
The product team had invested heavily in MoEngage but wasn't seeing results. D30 retention sat at 31%, push opt-out rates were climbing to 34%, and the team had no clear read on what was working or why. The core problem wasn't the platform, it was the architecture.
How we did it
Map what actually exists
We spent the first two weeks pulling every live campaign, every event definition, every segment rule, and every journey into a spreadsheet. The pattern was immediately clear: journeys were built around message types, not user states. There were 40+ campaigns with overlapping audiences, no suppression logic, and events that fired inconsistently across platforms. The MoEngage instance was technically live but functionally broken.
Rebuild from user behaviour up
We rebuilt the event taxonomy from scratch: 22 clean events with consistent naming, proper property schemas, and cross-platform QA. Alongside this, we designed a segmentation model based on user lifecycle stage (Activated, Engaged, At-Risk, Dormant) with concrete entry/exit criteria. Every journey was mapped against lifecycle stages before a single message was written.
Ship and measure
With the foundation in place, we shipped 12 journeys across onboarding, feature adoption, re-engagement, and retention, each with control groups and clear success metrics. We ran weekly performance reviews, killed what wasn't working, and doubled down on sequences that moved D30 numbers. By day 90, the data was clear.
“Faith took over our MoEngage instance when it was basically a liability. Within 6 weeks we had proper segmentation, 8 live journeys, and our D30 retention was climbing for the first time in a year.”
Want results like these?
Let's talk about what's broken in your lifecycle and what it would take to fix it.