A mobile measurement partner, or MMP, is an independent company whose software decides which ad, if any, led to each app install and the in-app events that followed. It sits between the app, the ad networks and the advertiser, and its job is to give one consistent answer to the question every user acquisition team asks: what did we get for what we spent, by network, campaign and creative.
Without one, each network reports its own installs by its own rules, and the totals add up to more installs than the app actually received, because several networks claim the same user. The MMP is the referee.
What an MMP does
- Attribution: matches each install, re-install or re-engagement to the touchpoint that earned it, by a consistent set of rules.
- Event measurement: records in-app events — tutorial complete, level reached, purchase, subscription — and ties them back to the same source.
- Postbacks: sends installs and events back to the networks, so their bidding algorithms can optimise toward users who actually do something.
- Cost and revenue aggregation: pulls spend from networks and revenue from the app to report return on ad spend by campaign.
- Fraud protection: filters installs and clicks that look manufactured, such as click spamming, click injection and device farms.
- Deep linking: routes a click to a specific place in the app, and through the store first if the app is not installed.
How attribution works
Tracking links and SDK
Most networks report clicks and impressions to the MMP through a tracking link or a server-to-server integration, carrying details such as the campaign, the ad and the creative. Inside the app, the MMP's SDK reports the first launch. The MMP then looks for touchpoints that could have caused that install and applies its rules to pick one.
Matching methods
Where a device identifier is available and the user has permitted it, matching is deterministic: the same advertising ID on the click and on the install. On Android, the Google Play Install Referrer passes the referring click's details to the app on first launch, which is also deterministic. Where neither is available, some MMPs fall back on probabilistic methods, which estimate a match from signals; on iOS, Apple's rules restrict fingerprinting, and MMPs have adapted how they use those methods. The details differ by provider and change with platform policy, so ask each MMP how it attributes on iOS today.
Attribution windows
An attribution window is how long after a touchpoint an install can still be credited to it. Click-through windows are typically measured in days and view-through windows in hours, with a click usually taking precedence over an impression. Defaults vary by MMP and by network, and many are configurable. Agree the windows with each network before comparing their numbers, because a network on a longer window will look better without being better.
Self-attributing networks
Some large platforms do not pass every click to the MMP. Instead, the MMP asks them whether they claim an install, and they answer. These are called self-attributing networks, and they include the biggest buyers of mobile traffic. The MMP still applies its own windows and de-duplication across sources, but it relies on the network's claim for the touchpoint itself.
iOS privacy: ATT, SKAdNetwork and AdAttributionKit
Since App Tracking Transparency, an iOS app must ask permission before it can access the device's advertising identifier for tracking, and many users decline. For those users, deterministic user-level attribution across apps is largely gone.
Apple's answer is privacy-preserving attribution. SKAdNetwork lets the App Store itself confirm that an install followed an ad, and sends a postback with campaign information and a limited conversion value, delayed and without user identity. AdAttributionKit is Apple's newer framework in the same family, which also supports alternative app marketplaces and re-engagement. MMPs help by managing the conversion value schema, decoding postbacks, and combining them with the user-level data that is still available. The SKAdNetwork and creative measurement guide covers what this means for testing creatives.
Creative-level reporting needs planning
Privacy-preserving postbacks carry limited detail, so creative-level attribution on iOS is coarser than it once was. Name creatives consistently, keep concepts few enough per campaign to read, and use Android and opted-in traffic to judge creative quality where iOS data is too thin.
The main MMPs
Four providers are most often compared. All four do attribution; they differ in emphasis, interface, integrations and commercial terms.
- AppsFlyer: one of the most widely used MMPs, with a very large catalogue of integrated partners, plus fraud protection, audience tools and privacy-focused measurement products.
- Adjust: a long-established MMP, part of AppLovin since 2021 while continuing to work with networks across the industry, known for attribution, fraud prevention and automation features.
- Singular: positions itself around combining attribution with cost aggregation and marketing analytics, so spend and return sit in one report.
- Branch: strongest known for deep linking and routing users across web, app and store, with mobile attribution alongside.
Others, such as Kochava, also serve the market. Pricing is negotiated and usually tied to volume; ask each provider directly.
How to choose an MMP
- List the networks you buy from today and plan to test, and confirm each is an integrated partner, including cost and creative-level data where you need it.
- Ask how the MMP attributes on iOS now — with and without ATT consent — and how it handles SKAdNetwork and AdAttributionKit conversion values.
- Check deep linking if your ads send people into specific content, offers or levels.
- Look at fraud protection: what it blocks, what it only flags, and how disputed installs are settled with networks.
- Check data export: raw data to your own warehouse, API access, and how long data is kept.
- Review privacy and compliance: consent handling, data processing terms and where data is stored.
- Try the reporting with your own team. The tool people actually open daily is worth more than the one with the longest feature list.
Common setup mistakes
Most disagreements between an MMP dashboard and a network dashboard come from setup rather than from either side being wrong. Check these before escalating a discrepancy:
- Different attribution windows on the two sides, so each counts a different set of installs.
- Time zones that do not match, which moves installs across day boundaries in every daily report.
- Events sent from the app but not mapped to the network's postback, so the network optimises toward installs only.
- Inconsistent campaign and creative names across networks, which makes cross-network reporting impossible to read.
- An SDK integrated without testing on a real device, so the first launch is never reported for one platform.
- Comparing attributed installs with the store console's total, which also counts organic installs.
Where creative and the MMP meet
An MMP is where creative testing becomes measurable. It tells you which creative produced installs, and then which produced users who stayed and paid — the comparison that decides whether a playable is worth more than a video. For that to work, creative names must be consistent across networks, and the click-through from the ad must carry the MMP's parameters into the store and the app.
Playable Ads Maker supports that click-through. Its exports can open an MMP tracking link or a deep link with campaign parameters attached, so the touchpoint in a playable or interactive end card is recorded like any other click. The deep linking and dynamic CTA guide explains how to set it up, and playable ad metrics explains how to read what comes back.
Create playable end cards in minutes—no code required.
Open Playable Ads Maker