Short answer: each ad network counts every install its ads touched, by its own rules, and several networks can claim the same install. Your MMP gives each install to one source, using its own evidence and windows. So network totals usually add up to more than your real installs. Compare like with like (same dates, same windows, clicks only), and use the MMP's de-duplicated count as your total.
What are self-attributing networks?
Large platforms such as Meta, Google, TikTok and Snapchat decide on their own side whether their ads drove an install. They are called self-attributing networks, or SANs. Instead of relying on outside click tracking, they receive install and event data and match it against what their users saw and clicked.
Why do the numbers differ?
- Everyone claims their share. If someone saw a TikTok ad, clicked a Meta ad and then searched the store, both networks may count the install. Your MMP credits it once.
- Views count too. Networks can count an install after someone only saw an ad, without clicking. An MMP that credits clicks will not.
- Different windows. A network may count installs for days after a click or a view; your MMP may use a different window. The longer the window, the more installs it claims.
- Different dates. Networks often report by the day of the click or the impression; MMPs usually report by the day of the install. A campaign that ends on the 31st keeps gaining installs in the MMP for days.
- Modelled conversions. Some networks add estimated conversions for people they cannot observe directly. Those never appear in an MMP's install-level data.
- Reinstalls. A network may count a reinstall as a new install where an MMP does not.
- Fraud. An MMP may reject or tag installs that a network still counts.
How do I compare them fairly?
- Use the same date basis: install date in both, if the network allows it.
- Use the same window, and compare installs from clicks only.
- Compare one network at a time, against the MMP's installs for that network.
- Expect the network to be higher. A gap that stays steady over weeks is normal; a gap that suddenly widens deserves a look.
- Use each number for its job: the MMP's for totals and for splitting budget between networks, each network's own for optimising inside that network.
What should your MMP show when they disagree?
The evidence. For any install you should be able to see which signal matched, how sure the match is, and which other clicks, from which networks, lost and why. Without that, "the MMP says 900 and Meta says 1,200" is an argument nobody can settle.
What does Measura do?
- Clicks, not views. Measura credits installs to clicks, and does not count ad views.
- Meta on Android. Measura reads Meta's encrypted install referrer once you add your decryption key, so installs from Meta ads are credited to the exact ad, from the Play Store's own record. Our Install Referrer guide explains how.
- Google's app campaigns. Once you connect Google Ads, Measura sends each install to Google's app conversion API, and Google answers whether its ad drove it. Measura's own evidence wins: Google's claim moves an install to Google only when Measura found no click for it, and the claim is kept on the install's record either way.
- Every install explained. Each install carries its confidence score, the signal that matched and the clicks that lost, on every plan. The full rules are in the methodology.
Questions people ask
Which number should I give my finance team?
The MMP's de-duplicated installs, because each install is counted once. Quote a network's numbers only as that network's own view of its results.
Is my MMP undercounting?
It can, if some campaigns run without tracked links or the SDK is set up wrongly, so check both. But a network reporting more than your MMP is expected, and is not by itself a sign of a problem.
Does this apply to in-app events and revenue too?
Yes, for the same reasons: each network counts the purchases it thinks it influenced. An MMP ties each purchase to the install's one source.