Attribute every app install to the campaign that caused it

Why don't Meta, TikTok and the app store report the same install numbers?

Meta, TikTok and the app store count installs differently, and the ad networks' combined totals usually exceed the real number of installs. A self-reporting network is an ad platform — such as Meta, Google or TikTok — that measures and reports its own campaign results, so its numbers cannot be independently verified from the platform itself.

Each self-reporting network credits itself with every install it can connect to one of its own ads, so an install that followed an ad on Meta and an ad on TikTok is claimed by both. An illustrative example: network A claims 100 installs, network B claims 80, and the store records 140 installs over the same period — 40 installs were claimed twice.

A mobile measurement partner (MMP) is a third-party platform that attributes mobile app installs and in-app events to the advertising campaigns that caused them, and reconciles those counts across ad networks that each self-report. An MMP assigns each install to a single source, so per-network totals add up to the real total instead of exceeding it.

Install attribution is last-touch bookkeeping, not a measure of causality. Attribution says which ad preceded an install; incrementality testing, a separate discipline, asks whether the install would have happened without the ad. Traqio does attribution and does not claim to measure incrementality.

How are app installs attributed on Android?

Deterministic attribution matches a click to an install using an identifier that is present on both sides, so the match is exact rather than inferred. On Android, Traqio's identifier is a click id carried from the tracking link through the Play Store by the Google Play Install Referrer.

The Google Play Install Referrer API returns the referrer string attached to a Play Store listing visit, together with click and install timestamps, letting an app read on first launch which ad click preceded its own install. Google's documentation states that the install referrer information is available for 90 days and does not change unless the app is reinstalled.

The Android round trip has three steps. A Traqio tracking link records the click and its click id; the Play Store keeps the click id in the referrer string; the Traqio SDK reads the referrer on first launch and the Traqio server matches the click id to exactly one recorded click. No advertising identifier is read at any step.

An Android install with no matching click inside the window is recorded as organic. Organic installs are observed but never billed as conversions.

How are app installs attributed on iOS after ATT?

SKAdNetwork (SKAN) is Apple's attribution framework: instead of user-level data, Apple sends the ad network a signed, delayed and aggregated postback stating that a campaign produced an install, plus a small conversion value describing early in-app behaviour. Traqio attributes iOS installs from those postbacks.

Traqio verifies Apple's signature on every SKAN postback, decodes the conversion value against a versioned mapping defined per app, and stores the raw postback untouched. Postbacks from SKAdNetwork 2.1 and later are accepted; unsigned or invalid postbacks are rejected.

SKAN attribution is campaign-level, not user-level. Apple's documentation states that the first postback arrives after a random 24–48 hour delay once its conversion window closes, and that Apple limits the detail in each postback according to a crowd-size tier. As a result, per-user iOS ROAS does not exist in Traqio, and Traqio does not model one into existence.

App Tracking Transparency (ATT) is the iOS framework, mandatory since iOS 14.5, that requires an app to ask the user's permission before accessing the IDFA or tracking them across other companies' apps and websites. Traqio does not need that permission, because SKAN attribution does not depend on it.

What Traqio refuses to do

Traqio does no probabilistic attribution, no fingerprinting and no clipboard reads, on any platform. Probabilistic attribution estimates which click caused an install by comparing signals such as IP address, device model and timestamp, without a shared identifier — which is fingerprinting.

Apple's user privacy guidance answers the question “Can I fingerprint or use signals from the device to try to identify the device or a user?” with “No”, and states that apps referencing SDKs — attribution services included — that engage in the practice may be rejected from the App Store. Traqio's SDK therefore collects neither the IDFA nor Android's advertising id (GAID), and never reads the pasteboard, whose paste prompt degrades the user's first launch.

The refusal has a cost that is worth naming: an install that no deterministic signal connects to a click is counted as organic rather than guessed at. Traqio prefers an undercount it can explain to an overcount it cannot.

What attribution window does Traqio use?

An attribution window is the maximum time between an ad click and an install for the click to receive credit. Traqio uses a 7-day click-through window: a click stays eligible for 168 hours, and every match inside it is a click id, never an inference.

The 7-day window is set for the whole Traqio deployment and applies to every customer identically; it is not configurable per app today. Traqio does not do view-through attribution — an ad impression without a click never receives credit for an Android install.

On iOS the window is Apple's, not Traqio's. Apple's SKAdNetwork documentation gives a user 30 days to install after a StoreKit-rendered ad and 24 hours after a view-through ad, and SKAN applies those rules before any postback reaches Traqio.

What you do to get attributed installs

Getting attributed installs in Traqio takes three steps: add the Traqio SDK to the app, route campaign links through a Traqio tracking link, and connect the Meta and TikTok ad accounts whose spend should be joined to the installs.

Every step above is available on the free plan, which covers 1,000 conversions a month. None of the mechanisms on this page is reserved for a higher tier, because Traqio's plans differ in billed volume only.

• The React Native / Expo SDK is the packaged, installable SDK today. Native Android, native iOS, Flutter, Unity, Capacitor, Kotlin Multiplatform and .NET MAUI SDKs exist as source scaffolds and are not published packages yet.
• iOS SKAN postback handling is live on the server side; the app declares Traqio's postback endpoint and updates its conversion value.
• Connecting Meta and TikTok is what adds spend, and therefore ROAS; attribution itself works before any ad account is connected.

Where attribution still has gaps

Deterministic attribution covers traffic that passes through a Traqio tracking link, and that boundary matters on the two largest networks. Click and view attribution inside Meta's and TikTok's own app-install inventory requires being a listed measurement partner of each network, and Traqio is not listed yet (checked 2026-09).

Google App campaigns are a similar case: a Google App campaign cannot point at a third-party click tracker unless the tracker is a certified Google App Attribution Partner, which Traqio is not. Installs driven by those campaigns arrive in Traqio without a campaign attached.

On iOS, Traqio maps the SKAN ad-network ids of Meta and TikTok; postbacks from any other ad network are stored with their raw id under “other” rather than named.

When an enterprise MMP is the better choice

An enterprise MMP is the better choice when attribution has to cover more than two ad networks or more than mobile apps. Teams that need fraud protection, CTV or web attribution, deep linking, data clean rooms, incrementality testing or listed-partner status on every major network will find those in the established MMPs and not in Traqio.

Traqio fits the case where paid acquisition runs on Meta and TikTok, the question is which campaign produced which install, and the team wants the answer without a sales conversation. The comparison pages set out, source by source, what each established MMP publishes.