What is mobile app attribution?

What app attribution means

Mobile app attribution is the process of connecting each app install, and the actions that follow it, to the marketing touchpoint that caused it. A touchpoint might be a click on a paid ad, a view of a video ad, a tap on an influencer's link or a scan of a QR code on a poster.

Without attribution, a growth team sees a single total: installs went up this week. With attribution, the same team sees that one Meta campaign brought most of the new users, a Google campaign brought fewer but better ones, and the rest arrived with no ad at all. That second view is what lets you move budget with confidence.

Why is app attribution harder than web attribution?

Because the app store sits in the middle. On the web, a click lands on your own site and a cookie or URL parameter can carry the campaign details straight through. In apps, the click goes to the App Store or Google Play, the person installs, and only then does your app open for the first time with no built-in memory of the click.

Attribution therefore has to rebuild the link between two separate moments: the ad interaction before the store, and the first open after it. Chapter 2, how an install gets attributed, walks through that chain step by step.

Mobile platforms add a second layer of difficulty. Apple's App Tracking Transparency rules limit the device identifiers available on iOS, and Apple's App Store policy rules out fingerprinting a device to identify it. On Android, the Google Play Install Referrer gives a cleaner signal, but only for installs that come through Google Play.

Who does the attributing?

Several parties attribute the same install, and each one answers a slightly different question.

The large ad networks that report their own results are called self-attributing networks. They claim installs inside their own attribution windows, and they do not see what other networks did. That is covered in depth in self-attributing networks.

What an MMP adds

An MMP sits across all of these sources. It receives clicks and impressions from every network, reads the install signal from its SDK inside your app, and then picks one winner per install, usually the last eligible click. Because it de-duplicates across networks, its install total for a campaign is often lower than the total each network reports for itself.

The core building blocks

You will meet these terms in every chapter. Each one has a fuller entry in the glossary.

• Attribution window: how long after a click or view an install can still be credited to it.
• Click-through and view-through: whether the winning touchpoint was a click or only an impression.
• Last-touch (last-click) model: the most recent eligible touchpoint gets all the credit.
• Organic install: an install with no qualifying touchpoint inside any window.
• SDK: the code library inside your app that reports installs and events.
• Postback: a message sent to an ad network or partner to report an install or event.
• Deep link: a link that opens a specific screen inside the app.

A worked example

Priya sees a video ad for a budgeting app on Instagram on Monday but does not tap it. On Wednesday she searches the Play Store, sees a Google App campaign ad, taps it and installs. On Thursday she signs up and links a bank account.

Here is how different systems might record her:

1. Meta may count a view-through install if its view window covers Monday to Wednesday and Meta's own rules allow it.
2. Google Ads counts a click-through install, because a Google ad was tapped and the install followed within its window.
3. The MMP sees both touchpoints. Under last-click rules, the Google click on Wednesday wins, so the install and the sign-up are credited to the Google campaign.
4. Your total installs in the MMP rise by one, not two.

Nobody is lying here. Each system applies its own window and its own view of the world, which is why the same week can show 100 installs from a campaign in one dashboard and 115 in another.

Why the numbers never match exactly

Differences between dashboards are expected and usually explainable. The most common causes are:

• Different windows: Google Ads, by default, counts first opens within 30 days of an ad interaction and in-app conversions within 90 days. An MMP may use a shorter window.
• Different definitions of an install: Google Play counts an install on download, while most MMPs count the first open.
• Different reporting dates: Google and TikTok report a conversion on the day of the ad interaction, while many MMPs report on the day of the install.
• Time zones: each ad account has its own reporting time zone.
• View-through credit: networks may count installs after an impression that an MMP gives to a later click elsewhere.
• Privacy limits on iOS: SKAdNetwork data arrives late, in aggregate, and sometimes without a conversion value.

What attribution is used for

Attribution data feeds several decisions that cost real money:

• Budget allocation: moving spend from campaigns with weak cost per install or poor return to ones that perform.
• Bid optimisation: sending installs and in-app events back to ad networks, so their algorithms find more people like your best users.
• Creative decisions: comparing which ad creatives bring users who stay and pay.
• Partner payouts: paying affiliates and influencers only for the installs they actually drove.
• Fraud checks: spotting patterns such as clicks that arrive impossibly close to an install.

The later chapters follow that path: events, postbacks and ROAS covers what happens after the install, and the final chapter explains how to compare MMPs if you need one.