Key takeaways
- iOS CAC usually rises after ATT because measurement broke, so the first fix is your SKAN or AdAttributionKit conversion value schema, not your bids.
- Map conversion values in AppsFlyer to the early in-app events that predict a real first order or registration, not to installs or app opens.
- Run bidding and budget decisions on MMP data and judge each channel on one shared metric: cost per new user who completes the key action.
- Rebalance Google App Campaigns, Apple Search Ads and Meta around new-user CAC, and give each change a full measurement cycle before you judge it.
If your iOS cost per new user doubled after Apple's App Tracking Transparency (ATT), the pattern is a familiar one. Many app teams we talk to in Indonesia describe it the same way: Android looks fine, iOS looks expensive, and every platform dashboard tells a different story.
In our experience the auction is rarely the main problem. The problem is measurement. When the signal that ad platforms optimise on is weak or wrong, they buy the wrong users, and you pay for it in CAC. Here is how we fix it, in the order we fix it.
Why iOS app CAC climbs after ATT
ATT asks every user whether an app may track them across other companies' apps. Most say no. For those users, ad networks lose the user-level link between an ad click and what happens inside your app.
Apple's answer is SKAdNetwork (SKAN), and now its successor, AdAttributionKit. Both report installs and post-install activity to ad networks in aggregated, delayed postbacks, with privacy thresholds that hide detail when volumes are small.
That has three practical effects:
- Signals arrive late. Platforms learn about a good user days after the click, not minutes.
- Signals are compressed. You choose what the app reports, through a limited set of conversion values. Choose badly and the platform optimises for the wrong thing.
- Every platform models the gap differently. Meta, Google and Apple each fill missing data with their own estimates, so their reported numbers overlap and do not add up.
The result is an account that looks efficient on dashboards while your real cost per new customer keeps rising.
Step 1: Get SKAN conversion values right in your MMP
The conversion value schema is the single most important setting in iOS app marketing, and it is often left on a default.
“Your iOS CAC is rarely too high. It is usually measured wrong.”
In an MMP like AppsFlyer, the SKAN conversion studio lets you decide which in-app events map to which values. Our rules of thumb:
- Map to the event that predicts value, not to the install. For a delivery or commerce app that is usually first order or add-to-cart. For a job app it is completed registration or first application.
- Keep the first postback simple and fast. The first window is where you get the most detailed (fine) value. Use it for the early event that correlates best with a real customer.
- Use coarse values for depth. Later postbacks only carry low, medium or high. Use them for repeat behaviour, such as a second order.
- Check volume against privacy thresholds. If a campaign is too small, Apple strips detail. Fewer, larger campaigns often report better than many tiny ones.
- Plan for AdAttributionKit. Apple positions it as the successor to SKAN and it works alongside SKAN today. Make sure your MMP SDK and schema are ready for both, so you are not rebuilding later.
Before you touch any of this, confirm the in-app events themselves fire correctly. We cover that in our tracking audit before scaling ads guide.
Step 2: Move bidding and reporting onto MMP data
Once the schema is right, stop running the account from platform dashboards.
Each platform's iOS numbers are partly modelled and each takes credit for the same users. Add them up and you can easily count more new users than you actually acquired. Decisions made on those numbers move budget toward whoever models most generously.
What we do instead:
- Make the MMP the source of truth. All budget, bid and pause decisions use AppsFlyer data, deduplicated across channels.
- Send the same optimisation event to every platform. Google App Campaigns, Meta and Apple Search Ads should all optimise toward the same in-app event from the same schema.
- Set targets from MMP CAC, not platform CPA. If AppsFlyer says your true cost per new ordering user is X, your tCPA or bid caps should be built from X, adjusted for each channel's reporting lag.
- Keep platform data for diagnosis only. It is still useful for creative, placement and query analysis. It just should not decide the budget.
Step 3: Rebalance the channel mix around new-user CAC
With trustworthy data you can finally compare channels fairly. Each one plays a different role on iOS.
Google App Campaigns reach users across Google Search, YouTube, Discover and the Display network. On iOS they rely on Google's on-device and SKAN-based measurement, so they reward a clean conversion value schema and consistent conversion events. When the signal is right, they scale well.
Apple Search Ads (now branded Apple Ads) catches users already searching in the App Store. Intent is high, and Apple's own attribution does not depend on ATT consent the way third-party networks do. Track Apple Search Ads ROAS and new-user CAC by keyword group: brand, category and competitor terms behave very differently.
Meta is strongest for demand creation and creative testing. On iOS it depends heavily on your SKAN setup and on Meta's own modelled reporting, so a weak schema tends to hurt Meta first. It earns its share when creative is strong and the optimisation event is a real customer action.
How we shift the mix:
- Rank channels and campaigns by new-user CAC from the MMP, not by installs or platform CPA.
- Move budget in steps toward the lowest CAC, keeping each channel above the volume it needs to report clearly.
- Protect demand-creation channels from being cut purely on last-touch numbers. Check whether branded search and Apple Search Ads brand terms drop when you pull back.
- Repeat monthly, not daily. SKAN data needs time to settle.
Step 4: Judge success on new-user CAC
Installs are cheap to buy and easy to inflate. App install CAC on its own tells you very little. The metric that matters is cost per new user who completes the key action, ideally split by platform and channel.
A simple scorecard we use:
- New-user CAC (MMP, deduplicated) as the headline number
- Channel ROAS from the MMP for revenue-driven apps
- New user volume, so CAC improvements are not just the result of cutting reach
- Signal health: share of postbacks with usable conversion values, and campaigns below privacy thresholds
Give each structural change at least one full postback cycle, and ideally a few weeks, before you call it.
What we did for HappyFresh
HappyFresh is a grocery delivery app in Indonesia. iOS acquisition had become expensive and every platform claimed a different result. We worked across Meta, Google App Campaigns, Apple Search Ads and AppsFlyer in three moves:
- Rebuilt iOS measurement. We redesigned the SKAN conversion value setup in AppsFlyer so early postbacks reflected the actions that lead to a real first order.
- Moved bidding onto MMP data. Targets and budget decisions came from AppsFlyer, not from each platform's own reporting.
- Reset the channel mix around new-user CAC. Budget moved toward the campaigns that produced new ordering users at the lowest true cost.
The results:
- -69% new-user CAC in 3 months
- 17.8x Google iOS ROAS
- 10.2x Apple Search Ads ROAS, up from 7.5x
- +94% new users in a record month
The same principle works outside iOS commerce. For Pintarnya, a job app, the work centred on Meta and registrations, with -68% Meta cost per registration and 3.6x registrations. The full iOS case is in the HappyFresh story.
Where to start this week
If you only do three things:
- Export your current SKAN conversion value schema and check whether it maps to installs or to real customer actions.
- Compare platform-reported iOS conversions with your MMP totals for the same period. The gap tells you how far off your bidding is.
- Rank every iOS campaign by MMP new-user CAC and move one step of budget toward the best.
Paid acquisition gets cheaper when people already know your name, so it also pays to get your brand cited in AI search. For hands-on help, explore our performance ads and tracking services.
Want us to review your iOS measurement and channel mix? Book a call and we will tell you where your CAC is leaking and what we would fix first.
Questions we get asked
Why did my iOS app CAC go up after ATT?
Because most iOS users decline tracking, ad platforms lost user-level conversion data and now model results from delayed, aggregated postbacks. Campaigns optimise on weaker signals, so they buy cheaper but lower-quality installs. CAC rises even when the auction did not get more expensive. Rebuilding SKAN conversion values in your MMP restores a usable signal.
What are SKAN conversion values?
Conversion values are numbers your app reports to Apple after an install, describing what the user did. Apple sends them to the ad network in delayed postbacks. Since SKAN 4 there are up to three postbacks across a longer window, with fine values in the first one and coarse values (low, medium, high) in later ones, depending on privacy thresholds.
Should I use AppsFlyer or the ad platform numbers for iOS?
Use your MMP, such as AppsFlyer, as the single source of truth for decisions and keep platform dashboards for diagnosis. Each platform counts its own conversions with its own modelling, so their totals overlap and overstate results. An MMP deduplicates across channels and applies one conversion value schema to everyone.
Is Apple Search Ads worth it for apps in Indonesia?
Usually yes, for apps with clear search demand. Apple Search Ads catches users who are already looking for your category in the App Store, and its attribution does not depend on ATT consent the way other networks do. Keep it on brand and category terms first, then expand only where new-user CAC holds.
About FOXA DigitalFOXA Digital is a performance marketing agency in Jakarta, Indonesia. Since 2018 we have run paid media, tracking, SEO and content for apps, consumer brands, hospitality and attractions, judged on CAC, ROAS and leads.
Want us to look at your numbers?
30 minutes. No deck. You leave with three things to fix this week.



