Measuring app growth after ATT: what Gulf CMOs should actually track

Ask five growth leads in Dubai or Riyadh how their iOS campaigns are performing and you will get five different kinds of answer. One quotes platform-reported ROAS, one quotes an MMP dashboard, one quotes blended cost per order, and two will quietly admit they aren't sure. None of them are wrong. iOS measurement simply stopped being one number in 2021, and many reporting decks never caught up.

This is a plain-English guide to what changed, what is available today, and what a CMO in the Gulf should ask their team to report.

What changed, in one paragraph

Since iOS 14.5, apps must ask permission through App Tracking Transparency (ATT) before they use the device's advertising identifier (IDFA) to track people across other companies' apps and websites. Most people, in most markets, say no. Without that permission, the old model of matching every ad click to every install and purchase at user level no longer works on iOS for the majority of your traffic.

The three sources of truth you now live with

Source	What it gives you	Where it falls short
Consented, user-level data	Full-fidelity attribution for users who opted in, plus all of your own first-party in-app data	Only a slice of iOS users, and that slice is not representative
Apple's frameworks: SKAdNetwork and AdAttributionKit	Privacy-preserving install and conversion postbacks for every eligible user, consented or not	Delayed, aggregated and capped in detail, especially for small campaigns
Modelling: MMM, incrementality tests, platform models	Answers to "what did this budget actually cause?"	Needs volume, discipline and time, and is only as good as its inputs

SKAdNetwork (SKAN) is Apple's original framework. Its fourth version added multiple postbacks over longer windows and "coarse" values for smaller campaigns. AdAttributionKit, introduced with iOS 17.4 in 2024, is its successor. It is interoperable with SKAN and Apple has kept extending it, for example to re-engagement. For a CMO the point is simple. Apple will tell you, with a delay and with less detail, which campaigns drove installs and early conversions. It will not tell you who those people were.

On Android, the advertising ID is still available, but plan as if it won't be forever. Teams that build good habits on iOS are better prepared for whatever Android does next.

Design the conversion value like it is a product decision

The most important measurement decision your team makes is one most CMOs never see: the conversion-value schema. This is the small set of values your app reports back through SKAN or AdAttributionKit in the first hours and days after install. If it encodes "opened the app", you will optimise for people who open the app. If it encodes a signal that predicts revenue, such as a first order, a completed KYC, a subscription trial or a deposit, your campaigns can learn to buy customers rather than installs.

Questions to ask your team:

• Which early action best predicts 30- or 90-day revenue in our data, and how strongly?
• Does our schema still make sense in Ramadan, when behaviour and time-of-day patterns shift?
• When did we last revisit it? A schema designed at launch is rarely right two years later.

What to report, and to whom

One dashboard for everyone is how measurement debates start. Split it instead:

• Board and CFO: blended efficiency. Total marketing spend against new customers, revenue or contribution margin, by month and by market (UAE, KSA, and so on). It is hard to argue with because it doesn't depend on attribution at all.
• Marketing leadership: channel contribution. Blend user-level, SKAN/AdAttributionKit and platform data into a single view, with the methodology written down. Change the method rarely and announce it when you do.
• Channel owners: fast signals. Platform metrics and early postbacks for day-to-day optimisation, with the explicit understanding that these are steering signals, not the scoreboard.

The Gulf-specific wrinkles

• iOS often punches above its weight. In many Gulf categories iOS users are a disproportionate share of revenue, so the iOS measurement gap matters more than device share alone suggests.
• Bilingual campaigns split your data. Arabic and English versions of the same campaign can each fall below the privacy thresholds that unlock detailed postbacks. Fewer, larger campaigns often measure better than many small ones.
• Seasonality breaks comparisons. Ramadan moves about eleven days earlier each year in the Gregorian calendar, so "versus same month last year" is often meaningless. Compare like-for-like seasons instead.
• Web-to-app is common. Many Gulf brands acquire on the web and convert in the app (or the reverse). Make sure links, deep links and web attribution are stitched together, or the app will look worse than it is.

A checklist for your next quarterly review

1. We can explain, in one slide, how iOS performance is measured and what is modelled.
2. Our conversion-value schema maps to a revenue-predictive event and was reviewed in the last six months.
3. The board sees blended efficiency by market; the team sees channel-level signals.
4. We ran at least one incrementality test this half and changed a budget because of it.
5. Arabic and English campaigns are structured so that neither falls below privacy thresholds.
6. Web and app journeys are attributed together.

Disclosure: Irfan Shaik works at Singular, a marketing measurement company. This piece is written to be vendor-neutral. It describes how the measurement landscape works, not any one product.