AEM vs App Store Connect off by ~30%, CPA blows up when scaling. Signal issue?

Hi guys,
I'm running iOS app campaigns for a subscription app on Meta with AEM (no SKAN set up) for a month now & I notice that AEM installs & subscribe events recorded on meta are ~30% below actual App Store Connect numbers.

When I tried scaling by duplicating a well-performing campaign (with a big budget) , CPA spiked hard, x3 our target CPA.
My guess is the missing signal that cause this happens, but not sure if that's the real cause or if I'm missing something else entirely.

Anyone dealt with something similar? What ended up being the actual problem for you?

Underreporting with AEM is normal and expected, and you're surely missing conversions because Meta doesn't track view-through conversions under AEM (same limitation ASC has too).

There's no way to fully close that gap. What most apps are running now is EAC campaigns using Appstack (or web-to-app using an MMP), so you can run a web sales campaign and get view attribution. That way, the algorithm receives more signals.

This will cause the opposite situation, though: you'll likely see overreporting in Meta vs ASC for the same reason: ASC doesn't count view conversions.

Thanks, davamos. Web-to-app is a solution I will definitely try. I also saw an article that recommends implementing CAPI for Meta campaign, would that bring more signal for a maximize conversion campaign? What are your thoughts on this?

web-to-app or EAC are both working with CAPI
If you meant working with a S2S connection instead of the usual MMP setup, that has also worked well for me!

Love it, thank you davamosk

Hi davamosk, my team has already implemented a CAPI flow from RevenueCat to Meta Ads. The impact on CPA (Purchase) is still inconclusive, as our product team is currently running a price A/B test, so we suspect that this may also be affecting our conversion rate.

However, after implementing CAPI, we did see a significant improvement in install performance, with CPI decreasing by around 30% so far, and the trend seems to be continuing ( new download with app refferer filtered also witnessed the same trend as well - no other change was made to the campaign around this time)

Can I attribute this improvement to CAPI, considering that the trend started the day right after we implemented it? I am still confuse since from what I have read, it would have a visible impact on cost per result ( cost per purchase event ) not cpi.