Self-Attributing Network

If you have ever run mobile user-acquisition campaigns and noticed that Meta, Google, and your dashboard each report a different number of installs for the same week, you have already met the self-attributing network. Understanding how SANs work is the single biggest step toward making your acquisition numbers trustworthy.

What a self-attributing network actually is

A self-attributing network, or SAN, is an advertising platform large enough that it refuses to hand over raw click-level data to outside measurement tools. Instead of letting your mobile measurement partner decide which click earned an install, the SAN performs the matching itself. When a conversion happens, your measurement partner asks each connected network, "Did you drive this user?" The SAN checks its own logs, and if it finds a matching click or view, it claims the credit and returns a yes.

Google Ads, Meta, TikTok, Apple Search Ads, and Snap all operate this way. They are walled gardens: they own enormous first-party login graphs, so they can match users across devices far better than any third party. The trade-off is transparency. You see the network's verdict, not the evidence behind it.

Why SAN numbers never match

Because each SAN grades its own homework, three structural mismatches appear in every account.

First, attribution windows differ. One network may claim a conversion up to seven days after a click; another counts view-through conversions within a day. The same install can legitimately be claimed by two networks at once.

Second, deduplication happens after the fact. Your measurement partner receives competing claims and applies a priority order to decide who wins. The network's own dashboard, however, reports every claim it made — including the ones your MMP later overruled. That is why platform-reported installs almost always exceed MMP-attributed installs.

Third, the SAN sees conversions your last-click model misses. View-through and cross-device matches inflate a network's self-reported tally relative to a stricter deterministic model.

How to manage SANs without losing your mind

The fix is not to make the numbers identical — that is impossible — but to pick one source of truth and reconcile against it deliberately.

Make your MMP the referee. Define a clear deduplication hierarchy so that when two SANs claim the same install, the same rule always decides. Document each network's click and view-through windows in a shared sheet before anyone compares performance, because comparing a 7-day-click network against a 1-day-view network is meaningless without that context. This is the same discipline behind sound multi-touch attribution: the model only works when everyone agrees on the rules.

Finally, reconcile weekly. Pull SAN-claimed installs, MMP-attributed installs, and store-reported installs side by side. A stable gap is fine and expected; a sudden divergence is a signal that a window changed, a SDK broke, or fraud crept in. For iOS specifically, layer this thinking on top of SKAdNetwork, whose aggregated postbacks add yet another, privacy-limited view of the same campaigns.

FAQ

Can I just turn off self-attribution and use my MMP's click data instead? For most SANs, no. They do not pass the raw click data your MMP would need to attribute independently, so you accept their claimed conversions and dedupe afterward. That is the deal you make to advertise inside a walled garden.

Why does my SAN report more installs than my MMP? The SAN reports every conversion it claimed; your MMP reports only the claims that won deduplication, plus it may use a stricter window. The platform number is therefore almost always higher.

Which number should I report to finance? Use your MMP as the single source of truth so every channel is judged on the same rules. Treat SAN dashboards as optimization signals for that specific network, not as your company-wide total. Tools like AppsFlyer, Adjust, and Singular exist precisely to give you that neutral referee.