AppsFlyer Attribution, Explained

AppsFlyer is a mobile measurement partner (MMP): the impartial referee that attributes app installs and in-app events to the campaigns that drove them, across every ad network you buy from. Its attribution toolkit spans deterministic matching where identifiers are available and consented, probabilistic modeling where they aren’t, Apple’s SKAdNetwork on iOS, view-through crediting, and attribution for web, CTV and beyond.

This guide is the plain-English tour: what an MMP does, how each attribution method works, the listed pricing, and where MMP data fits in a wider measurement stack. Claims last verified August 2026, from AppsFlyer’s own site and glossary.

What an MMP is, and why apps need one

Using AppsFlyer’s own definition: an MMP attributes, collects and organizes app data to deliver uniform campaign metrics: an impartial referee ruling on attribution between ad networks. AppsFlyer states it serves 15,000+ brands, with a platform spanning measurement, deep linking, data collaboration and more, and attribution products covering mobile, web, CTV and PC/console.

Why the referee matters: without one, every network claims the installs it touched. Meta reports 100 installs, Google reports 80, and your app store shows 120 total: because both networks claimed some of the same users. An MMP applies one set of rules to the whole picture, assigns each install to one source, and produces numbers that sum to reality.

That’s the core value: one impartial scoreboard where every network otherwise keeps its own generous score.

How AppsFlyer install attribution works

The core model is last-touch: the final touchpoint before the install gets the credit. Attribution links carry campaign context; clicks and qualifying views count if the install lands within the configured attribution window.

Where exact device identifiers are available and consented, matching is deterministic: exact, individual-level connections between the touch and the install. One honest caution from our verification: specific default window settings vary and are configurable per account, so treat any “the window is X days” claim you read elsewhere with suspicion: the concept is what matters, and your account’s settings are what actually apply.

The privacy-era methods: SKAN and probabilistic

On iOS after App Tracking Transparency, the deterministic spine covers only consenting users, and two methods fill the gap.

SKAdNetwork (SKAN) is Apple’s privacy-centric framework: instead of user-level trails, ad networks receive aggregate conversion postbacks, delayed by at least 24 hours, carrying limited conversion values, with crowd-anonymity thresholds that suppress data from small campaigns. Campaign measurement survives; individual trails don’t. AppsFlyer’s tooling helps design conversion-value schemas that make the most of those constrained signals.

Probabilistic modeling estimates aggregate campaign performance statistically from signals like timing and device characteristics, explicitly without individual identifiers: an estimate with stated uncertainty rather than a followed trail.

The practical upshot: mobile attribution is now a mixed-methods system: deterministic where consented, modeled where not, aggregate where Apple says so. That’s the honest shape of the whole privacy era, covered more broadly in cookieless and ATT-era attribution.

Pricing and practical notes

As listed in August 2026: a free Zero plan for owned-media-only apps (AppsFlyer itself notes it’s not suitable for apps running paid campaigns); Growth is pay-as-you-go, priced at 0.07 dollars per conversion after a welcome package; Enterprise is custom.

Two practical notes for buyers: conversion-based pricing scales with volume, so model your expected conversion counts before committing: high-volume apps should do that arithmetic early. And reattribution: re-engaging lapsed users who reinstall or return, is its own measurement stream, worth configuring deliberately rather than discovering in an invoice.

Where MMP data fits in the measurement stack

The honest boundary: an MMP referees install credit between networks superbly, and that’s a different job from measuring incrementality. Last-touch credit inside the mobile ecosystem can’t tell you whether those installs would have happened anyway, how mobile spend interacts with web and offline channels, or where the next euro of budget should go across the whole mix.

MarSci helps many app-first business identify the true value of each marketing channel without the need for analysing individuals. You can have an accurate understanding how users generate discover your brand: MarSci’s data-driven attribution solutions.

The sports version: the MMP tells you who scored each goal; the aggregate models tell you how the team actually wins games. App businesses at scale run both: the referee’s whistle for network-level decisions, the league table for the budget. If your measurement stops at the whistle, book a call with MarSci.

Frequently Asked Questions