What is mobile attribution?

Mobile attribution is the process of connecting an app install or in-app event to the ad, campaign or channel that caused it, so marketers can see which spend drives users and revenue.

A mobile measurement partner is a company that collects install and event data from an SDK in your app and matches it with data from many ad networks. Examples are AppsFlyer, Adjust, Branch, Kochava and Singular.

How does attribution work on iOS without the IDFA?

For users who don't allow tracking under ATT, attribution relies on Apple's SKAdNetwork and AdAttributionKit, which send aggregated, delayed reports per campaign instead of user-level data.

What is an attribution window?

The period after an ad click or view during which an install is still credited to that ad, for example 7 days after a click. Different windows produce different numbers, so keep them consistent.

Mobile Attribution: How It Works, Models & MMPs Explained

How mobile attribution works

A user sees or taps an ad on an ad network, a social platform or a website.
The network records the engagement (click or impression) and reports it.
The user installs the app and opens it. An SDK in the app records the install and later events, such as sign-up, purchase or subscription.
The attribution provider matches the two and credits the campaign that led to the install.
Later events are attributed to the same source, so you can see revenue per campaign, not just installs.

The companies that do this matching across many ad networks are called mobile measurement partners (MMPs). Well-known ones are AppsFlyer, Adjust, Branch, Kochava and Singular. Large platforms such as Meta, Google, TikTok and Apple Search Ads report their own results and are called self-attributing networks.

Attribution models and windows

Last-click (last-touch): the last ad the user tapped before installing gets the credit. The industry default.
View-through: an install can be credited to an ad that was seen but not tapped, usually within a short window.
Attribution window: how long after a click or view an install still counts, for example 7 days after a click and 24 hours after a view.
Re-attribution: returning users who reinstall or come back after a long break can be attributed to a re-engagement campaign.

The model and windows you choose change the numbers, so keep them consistent when you compare campaigns.

Mobile attribution on iOS after ATT

Since App Tracking Transparency (ATT), apps must ask permission before using the advertising identifier (IDFA) for tracking. Many users decline. For them, user-level attribution isn't possible, and iOS attribution relies on Apple's privacy-preserving frameworks, SKAdNetwork and AdAttributionKit. These send aggregated, delayed reports (postbacks) per campaign, with limited detail about what users did after installing.

In practice, iOS teams combine three sources: the privacy-preserving postbacks, consented user-level data, and the platforms' own reporting.

Mobile attribution on Android

On Android, the Google Play Install Referrer passes information about where an install came from directly to the app. Combined with the advertising ID (where the user hasn't opted out), Android attribution remains more detailed than iOS.

Why mobile attribution matters

Without attribution you can't calculate cost per install, compare campaigns against lifetime value, or decide where the next marketing dollar goes. It's the basis of every user acquisition budget decision. It also powers deferred deep linking, which sends new users to the right screen after install.

Limits of attribution

Organic and word of mouth are undercounted. Installs that came from a friend, a review or search look "organic"; see organic installs.
Last-click over-credits the final touch. A user who saw five ads credits only the last one.
Incrementality is a separate question. Attribution says which ad was last; it doesn't prove the install wouldn't have happened anyway. Holdout tests answer that.

Example

A team runs ads on three networks. Attribution shows network B has the highest CPI, but its users subscribe twice as often. Measured against 90-day revenue, B is the cheapest channel, so the team shifts budget to it.