How we scaled Meta trials without the budget Meta wanted.
Meta said we needed $250/day to optimise. We found a better way.
Meta AdsWhat was broken
Meta required $250/day to optimise for trials
To exit the learning phase on the start_trial event, Meta needs 50 conversions per week. That required a daily budget the client could not sustain.
Low-budget trial campaigns produced inconsistent results
Running below Meta's threshold meant the algorithm never learned. Results were unpredictable and cost-per-trial was too high to scale.
No sustainable trial acquisition strategy at current budget
The business needed a repeatable, profitable source of trial users — without doubling the media spend to get there.
What we did
Identified a higher-volume proxy event
Found that the thirty_second_session event occurred far more frequently than start_trial — and that most users who hit 30 seconds had already reached the paywall.
Shifted campaign optimisation to the proxy event
Restructured all campaigns to optimise for thirty_second_session instead of start_trial. More events, faster learning, lower cost.
Fed Meta's algorithm enough signal to work
Higher event volume allowed the algorithm to exit the learning phase, improve targeting accuracy, and bring in more trial-ready users consistently.
Adjusted targeting around the new signal
Refined audience parameters to focus on users most likely to generate the proxy event — building a more efficient, scalable campaign structure.
What we achieved
By giving Meta a smarter signal — not a bigger budget
Campaigns exited the learning phase. Trial volume increased. Cost per trial dropped. All without increasing daily spend above what the budget allowed.
Learning phase completed
The algorithm had enough conversion data to optimise effectively — something impossible when targeting start_trial at this budget level.
Budget increase required
The entire improvement came from optimising smarter — not spending more. The daily budget stayed the same throughout.
Why it worked
- Proxy events unlock low-budget optimisation — any event that correlates strongly with your target conversion can be used. The key is finding one with enough volume to feed the algorithm.
- 30 seconds on an app is meaningful intent — by that point, most users had navigated deep enough to reach the paywall. It was not a casual signal — it was a qualified one.
- Learning phase exit changes everything — once Meta's algorithm has enough data, targeting accuracy improves continuously. Below the threshold, it never gets there.
- The framework applies to any budget constraint — if your conversion event does not have enough volume, find one upstream that does. Map intent, not just actions.
Want results like this for your app?
Book a free strategy call. We will show you exactly what is holding your app back and what we would do about it.
Book a Free Strategy Call