Self-attributing networks
The biggest ad networks mark their own homework, and an MMP's job is to decide which of them actually gets the grade.
In short: A self-attributing network (SAN) is an ad platform that decides for itself whether its ads drove an install, then tells the MMP. Meta, Google and TikTok all work this way. Because each SAN only sees its own ads and uses its own windows, their totals are usually higher than the MMP's de-duplicated count.
What is a self-attributing network?
A self-attributing network is an advertising platform that matches installs to its own ads internally and reports the result, instead of sending every click to an outside tracker for matching. The MMP asks the network whether it has a qualifying touchpoint for an install, and the network answers yes or no, along with some campaign details.
Smaller ad networks usually work the other way. Their clicks go through an MMP tracking link, so the MMP sees the click directly and does the matching itself.
How does a SAN claim work?
The app's SDK reports an install to the MMP, the MMP queries each connected SAN, and each SAN replies with a claim if it saw a qualifying click or view inside its window. The MMP then applies its own rules, usually last click wins, to decide who gets the install.
Step by step:
- The person installs and opens the app, and the MMP SDK records the first open.
- The MMP sends the install details to Meta, Google, TikTok and any other connected SAN.
- Each SAN checks its own logs against its own attribution window.
- Each SAN that finds a match returns a claim.
- The MMP compares all claims and its own tracked clicks, and credits one source.
- Each SAN still counts the install in its own dashboard if it claimed it, whether or not it won in the MMP.
Step 6 is the root of most "the network shows more installs" questions.
Meta as a SAN
Meta counts app installs as installs recorded as app events and attributed to its ads. Meta says this requires the app to be registered with Meta, the SDK to be implemented correctly and the app to be connected to the ad account; otherwise installs can show as zero.
Points that matter for reconciliation:
- For app install campaigns, Meta's attribution is based on a 1-day, 7-day or 28-day click setting, and post-install events use 1-day or 7-day click.
- Meta has stated that it generally has a larger attribution window than most MMPs, on an older developer page.
- New Meta ad accounts default to Pacific time, and changing the time zone means a new ad account.
- On iOS 14 and later, Meta reports app campaigns through either its own Aggregated Event Measurement (AEM) or Apple's SKAdNetwork. AEM limits each app to 8 prioritised conversion events and may use statistical modelling where reporting is incomplete.
- Since 9 October 2024, Meta sends aggregated AEM reporting to MMPs, with identifiers removed and differential privacy added.
- MMP documentation reports that Meta re-enabled non-aggregated, last-touch data sharing with MMPs from 17 June 2025, for advertisers who accept its Advanced Mobile Measurement terms. This comes from MMP pages, not Meta's own terms page.
On Android, Meta also has its own Meta Install Referrer, which can carry view-through and multi-session click data. Chapter 7 covers it.
Google as a SAN
Google Ads only reports conversions when there was a Google Ads interaction from the same account. For app campaigns, Google counts first opens or installs within 30 days of the interaction by default, and in-app conversions within 90 days.
- Google Play installs are recorded on download. Installs from third-party Android stores are recorded on first open.
- Google Ads counts Android reinstalls by people who had uninstalled as install conversions.
- Conversions are reported on the day of the ad interaction, with separate "by conversion time" columns.
- Processing can take several hours, sometimes up to 24 hours, and some invalid conversions are removed later.
Google runs an App Attribution Partner programme for MMPs. Google's own page lists seven partners: Adjust, Airbridge, AppsFlyer, Branch, Kochava, Singular and Tenjin. Google also says discrepancies between Google Ads and third-party reports are expected, and recommends aligning conversion windows.
TikTok as a SAN
TikTok's help centre pages could not be loaded directly for this guide, so the following is reported from search-engine extracts of those pages:
- TikTok self-attributes when the click or impression falls inside its lookback window, while MMPs mostly use last click and de-duplicate across networks.
- By default, TikTok tracks installs within 7 days of a click or 1 day of a view, and in-app events within 90 days.
- Its window setting does not affect MMP reporting.
- Its ad account time zone is set when the account is created and cannot be changed.
- During the move to TikTok's SAN integration, numbers may fluctuate on the transition day.
Why SAN numbers differ from your MMP
| Cause | What happens | Effect |
|---|---|---|
| Overlapping claims | Two SANs claim the same install | Both count it; the MMP credits one |
| Longer windows | A SAN window covers an older click than the MMP's | SAN counts installs the MMP calls organic |
| View-through | A SAN claims after an impression | MMP may credit a later click elsewhere |
| Install definition | Google Play counts downloads | Platform total runs ahead of first opens |
| Reporting date | SAN reports by interaction day | Daily numbers differ, weekly totals closer |
| Time zone | Each account has its own time zone | Day boundaries shift |
| Modelled data | Meta AEM may model incomplete data | Platform totals include estimates |
A worked example
In one week, a fitness app gets 500 installs that the MMP attributes to paid media. Meta claims 320 installs and Google claims 260, a total of 580.
The gap of 80 comes from installs both networks claimed. For each of them, the MMP saw a click on one network and a click or view on the other, and gave the install to the most recent click. Meta and Google each kept their own count. The MMP's 500 is the de-duplicated figure; the networks' 580 is the sum of overlapping claims.
Reducing the gap
General practice for keeping SAN and MMP numbers close:
- Match each network's attribution window in the MMP.
- Compare on the same date basis and time zone.
- Filter to click-through only when comparing, to remove view-through differences.
- Confirm the integration is connected and the app ID is correct on both sides.
- Send the same events to both, mapped to the same names.
For more on how windows differ, see attribution windows, and for how iOS privacy rules change Meta's reporting, see iOS ATT and SKAN.
Questions on this chapter
Which networks are self-attributing?
The best-known are Meta, Google and TikTok, along with other large platforms that match installs internally. Smaller networks usually send clicks through an MMP tracking link instead.
Why does Meta claim installs my MMP gives to Google?
Meta only sees Meta ads, so it claims any install with a qualifying Meta touchpoint. If the person also clicked a Google ad later, the MMP gives the install to Google under last-click rules, but Meta still counts it.
Can I stop SANs from counting overlapping installs?
No, each SAN reports by its own rules. You can make comparisons fairer by aligning windows and using the MMP as the single cross-network total.
Does Google work with every MMP?
Google's App Attribution Partner page lists seven partners by name. Other MMPs may connect in other ways, so check how a specific vendor integrates with Google Ads.
Why did my TikTok numbers jump on one day?
TikTok's help centre, as reported in search extracts, says conversions may fluctuate on the day an app moves to its SAN integration. Annotate that date before reading trends across it.
What is mobile app attribution?
Mobile app attribution is the process of crediting each app install and in-app event to the ad, link or channel that drove it. It tells you which campaigns bring users, so you can move budget to the ones that work.