Meta · Published 2026-06-16 · 2 min read
Meta event mapping: StartTrial or Purchase?
When launching on Meta, choosing the optimisation event is the key decision. The whole performance of the campaign follows from it. The real choice is between two: optimising for Purchase, or optimising for StartTrial.
Optimising for Purchase
On Web2App, mapping app events onto Meta's standard events looks like this: install to Lead, trial to StartTrial, subscription started to Purchase, trial converted to Purchase.
Upsides
- Stable traffic. The algorithm learns on an actual charge.
- Simple economics. You look at a payer and you see the cost of that conversion, and it does not drift by geo, because it is real money rather than an intermediate event. If an annual subscription costs $60, then roughly speaking $60 is your revenue per user in Brazil and in the United States alike.
- Convenience. You can judge performance without leaving the ad account.
Downside
- Too few events. On a small budget, or with an expensive product, there are simply not enough payments to get the ad set through learning.
Optimising for StartTrial
Mapping: install to Lead, trial to Purchase, subscription started to Purchase. It is convenient to add subscription started alongside trials so that Meta has more events.
I do not pass trial converted, for two reasons: the information about that user is already in the trial event, and the KPI across trials plus subscriptions started is easier to read as ARPAS in Apphud or Adapty.
Upside
- Plenty of events. There are more trials than payments, so learning gets going faster — the minimum workable spend can be several times lower.
Downsides
- More work. Specific geos and specific creatives can show a low, sometimes absurdly low, trial-to-purchase rate. It needs constant monitoring of payback and conversion by geo and by creative.
- Conversion decays as learning progresses. To deliver more events, Meta increases their number at the expense of their quality, and trial-to-purchase gets worse over time.
Conclusion
- Purchase is the default: stable traffic and honest economics, without having to watch geo and creative.
- StartTrial is justified when the quality of the conversion into payment does not fall while you optimise on it. Then you get many events, fast learning and no loss of money. This is hard to predict in advance — it is settled by a test.
- Optimising for purchase buys expensive traffic; optimising for trial buys cheaper traffic. Which of the two adds up better in the end is, again, settled by a test.