Mobile attribution

Mobile attribution is the process of working out which ad, channel and campaign an app install, or an in-app action such as a sign-up or purchase, came from. The results show where your ad budget is buying users and at what price.

What is mobile attribution?

On a website, the ad link sends the user straight to the landing page, and cookies and URL parameters keep the trail. With an app, there is a stop at the App Store or Google Play between the click and the first app open, and advertisers cannot place tracking code there. Mobile attribution reconnects the two ends of that broken trail.

The job is usually handled by an MMP. An install that cannot be tied to any ad is counted as an organic install.

How installs are matched to ads

Install Referrer is a Google Play API that returns the referrer URL of the installed package and the click time. The data stays available for 90 days and does not change unless the app is reinstalled (Android Developers, Play Install Referrer Library, February 2026).

A person may see and click several ads before installing, so there has to be a rule for picking the winner. For SKAdNetwork and AdAttributionKit, Apple’s rule is that a click always beats a view, then the most recent one wins, with up to 6 impressions considered (Apple Developer, AdAttributionKit and SKAdNetwork interoperability, accessed October 2026).

What changed from 2021 to 2026

• April 2021: iOS 14.5 made App Tracking Transparency (ATT) mandatory. If the user does not agree, the IDFA is a string of zeros.
• Late 2021: when an Android user deletes their Advertising ID, apps receive only a string of zeros (Play Console Help, Advertising ID, accessed October 2026).
• October 2022: SKAdNetwork 4 on iOS 16.1 sends up to three postbacks per install.
• March 2024: AdAttributionKit launched with iOS 17.4, with support for alternative app marketplaces; re-engagement measurement was added in June 2024.
• October 2025: Google retired the Privacy Sandbox Attribution Reporting API on Android and Chrome (Google Privacy Sandbox, October 2025).

The upshot is that iOS and Android measurement now work very differently. Android can still match each install individually. On iOS, installs from people who decline ATT exist only as aggregate numbers.

Example

Hypothetically, a bubble tea chain launches an ordering app and spends 200 million VND, split evenly between iOS and Android. On Android, the MMP reports day by day how many installs and first orders each campaign brought in. On iOS, most results arrive through postbacks, and the first postback comes 24 to 48 hours after the first 2-day window closes. If the ad team shuts off the iOS campaign after 1 day because they “haven’t seen any orders yet,” they are making the call on data that has not arrived.

Mobile attribution vs. web attribution vs. attribution models

Web attribution solves the same problem for websites, using cookies and URL parameters. An attribution model is the rule for splitting credit between touchpoints, such as last click or data-driven. Mobile attribution can use many different models. The multi-touch attribution entry goes deeper into how credit gets divided.

Frequently asked questions

Why do install numbers in the MMP differ from App Store Connect or Google Play Console?

The stores count downloads. An MMP only sees an install once the app is opened and the SDK sends data back, so people who download but have not opened the app yet do not appear in the MMP.

Can mobile attribution measure people coming back to the app?

Yes. A return visit driven by an ad is called re-engagement. On iOS, AdAttributionKit can measure it, but only when the user clicks the ad.

To see how Adtek tracks app install campaigns on both iOS and Android, visit the App User Acquisition page.