Meta says you drove around 5k installs. Google claims credit for some of the same ones. Your MMP shows a lower total than both combined. None of them are lying, and a self-attributing network is why.
This is one of the most common sources of confusion in mobile measurement, and it is structural, not a bug. Once you understand how a self-attributing network reports, the mismatched dashboards stop being a mystery and become a reconciliation task with a clear answer for each cut. This post defines the entity, explains the mechanism, and gives you a rule for which source to trust when.
What Is a Self-Attributing Network?
Self-attributing network (SAN): an ad platform, such as Meta, Google, TikTok, Snap, or Apple Search Ads, that matches installs against its own click and view records and reports the conversions it believes it drove, rather than sending raw click data for a third-party measurement partner to attribute independently.
The defining trait is that a SAN both runs the ads and grades its own homework. It sees the impression, sees the click, and later receives a signal that an install happened, then decides whether to claim it.
The major self-attributing networks you will deal with:
-
Meta
-
Google (Ads and UAC)
-
TikTok
-
Snap
-
Apple Search Ads
A non-self-attributing partner works differently. It sends a click to your MMP and lets the MMP decide who won. That distinction is the root of every number that does not line up.
How SAN Attribution Actually Works
The flow is worth slowing down because it explains everything downstream.
-
A user sees and taps an ad inside the SAN's platform.
-
The SAN logs that click (and often the view) against the user or device.
-
The user installs and opens your app.
-
Your MMP notifies the SAN that an install occurred.
-
The SAN checks its own logs, and if it finds a matching click or view inside its window, it tells your MMP it should get the credit.
-
Your MMP records the SAN-claimed install rather than attributing it from raw click data.
Contrast that with a standard partner, where the MMP receives the click directly and runs its own last-touch logic. With a SAN, you are trusting the network's word that it touched the user first. That trust is reasonable, but it is also why two networks can both raise their hands for the same install.
Why the Numbers Never Match
Several independent factors stack up, and each one moves a number:
-
Different attribution windows. Each SAN applies its own click-through and view-through windows. A 7-day click window and a 1-day view window will count different installs than your MMP's configured window.
-
View-through credit. Some SANs claim installs from users who saw but never tapped an ad. Your MMP may weight or exclude view-through entirely.
-
Overlapping claims. When a user touched both a Meta ad and a Google ad, both networks can claim the install. Summing their dashboards double counts that user.
-
Counting and formatting differences. Currency conversion, timezone boundaries, and event mapping all shift totals at the margin.
If you want the platform-specific version of this for the most common offender, our guides on why your Meta ROAS does not match your MMP data walk through the exact settings to align.
Which Source to Trust for Which Decision
The mistake is treating this as a single "who is right" question. Different sources are authoritative for different jobs.
-
Trust the SAN for in-platform optimisation. Its creative, audience, and bidding signals are what train the algorithm. For tuning a Meta or Google campaign inside that platform, use the platform's own numbers.
-
Trust your MMP for cross-channel truth. Only the MMP sees all networks at once, de-duplicates overlapping claims, and ties installs to downstream revenue. For budget allocation across channels, the MMP is the referee.
-
Never blend the two without de-duplication. Adding SAN-reported installs to MMP-reported installs, or summing SANs against each other, inflates paid totals and quietly understates organic.
Our deeper guide on which ROAS number to trust covers the decision criteria in more detail.
How to Reconcile SAN and MMP Reporting
A repeatable reconciliation beats arguing about whose dashboard is correct.
-
Align attribution windows first. Set your MMP windows to match each SAN's click and view windows before comparing anything. Most "discrepancies" disappear here.
-
Decide your view-through policy. Choose whether view-through installs count, and apply it consistently across networks.
-
Use the MMP as the neutral referee. Let it de-duplicate installs that more than one SAN claims, and read total paid from there.
-
Validate per-network settings. Confirm postback and event mapping per SAN using the postback setup guide.
-
Pull one weekly cut. Network-reported installs next to MMP-attributed installs, with the gap explained, not just flagged.
Across multi-channel audits, the single most common reporting error we see is summing SAN-reported installs across Meta, Google, and TikTok without de-duplication, which inflates the apparent paid total and makes organic look weaker than it is. A single de-duplicated source removes that error entirely, which is the core argument in our piece on why attribution fails without a single source of truth.
How an MMP Sits Between You and the SANs
This is where the structure resolves. An MMP is the one neutral party that ingests every SAN's claim, applies one consistent set of windows, removes duplicate credit when two networks claim the same user, and connects the surviving install to revenue.
Platforms like Linkrunner do exactly this de-duplication across Meta, Google, TikTok, Snap, and Apple Search Ads in a single dashboard, so the marketer reads one cross-channel total instead of reconciling five self-reported ones. You can see how each network's signal is configured in the Meta Ads and equivalent setup docs. The point is not that the MMP overrules the platforms, it is that only the MMP can see across them at once.
FAQ
Which ad networks are self-attributing networks?
The main ones are Meta, Google, TikTok, Snap, and Apple Search Ads. They run the ads and report their own conversions, rather than handing raw clicks to a measurement partner to attribute.
Why do self-attributing networks report higher numbers than my MMP?
Because each SAN uses its own attribution windows and often claims view-through installs, and because two SANs can both claim the same user. Summed up, those overlapping claims exceed the de-duplicated total your MMP reports.
Can an MMP override a self-attributing network's claim?
An MMP does not usually overrule a SAN's self-reported credit, but it de-duplicates across all networks and applies one consistent window, so its cross-channel total is the trustworthy one for budget decisions.
How do I de-duplicate installs claimed by two networks?
Use your MMP as the single source that sees all networks at once. It assigns each install to one source under one consistent rule, which is impossible to do from separate platform dashboards.
Is view-through attribution why my SAN numbers are inflated?
Often, yes. View-through credit lets a SAN claim installs from users who only saw an ad. Decide your view-through policy and apply it consistently to bring the numbers closer.
Closing
Mismatched dashboards across Meta, Google, TikTok, and your MMP are not a sign that something is broken. They are the predictable result of self-attributing networks each reporting their own version of events. The fix is not to pick a favourite dashboard, it is to use each source for the job it is good at: the SAN for in-platform optimisation, the MMP for de-duplicated cross-channel truth.
If you want one reconciled view that de-duplicates SAN claims and ties them to revenue, request a Linkrunner demo. A good first step is to align your MMP attribution windows to each network this week, then pull a single side-by-side cut and watch most of the gap explain itself.