All work
AI EdTech · Conversion & Monetisation
Consumer AI Learning Platform
Using behavioural data and experimentation to move users from first value toward paid usage.
Context
An AI-assisted learning app that gave users strong first-time value, but struggled to convert free or trial users into paying subscribers. Usage was front-loaded around onboarding and dropped before the paywall.
The problem
- Users saw a strong first AI interaction, but did not experience enough repeated value before the trial ended.
- The upgrade moment was timed around the trial deadline, not around user readiness.
- No clear picture of which behaviours predicted subscription.
- Experimentation was ad-hoc; wins and losses were not tracked against revenue.
What I did
- Defined the behavioural signals that separated subscribers from churners.
- Redesigned the activation loop so users experienced meaningful AI value multiple times in the first week.
- Moved paywall prompts to moments of demonstrated value instead of calendar deadlines.
- Set up a structured test-and-learn process tied to conversion and revenue metrics.
Metric snapshot
Trial-to-paid conversion
4%11%
+7 pts
Week-1 repeat usage
31%52%
+21 pts
Experiments shipped / month
16
6x cadence
Outcome
- More trial users reaching repeated value before the paywall.
- Higher upgrade rate by prompting users at value moments instead of time limits.
- Clearer understanding of which product changes moved subscription revenue.
- A testing cadence the growth team could run without engineering for every experiment.
This work sits across the activation, retention and paid conversion services. For definitions and benchmarks behind these numbers, see the growth answers.