What an Attribution Discrepancy Is
An attribution discrepancy is the gap you see when two systems count the same thing differently. Meta Ads Manager says a campaign drove 1,200 installs. Your MMP says 950. Google Ads says your App campaign drove more than the MMP credits to it. Your finance team's revenue does not match the revenue in either.
Most discrepancies are not errors. Each system answers a slightly different question with its own rules. The work is to know which differences are expected, which point to a setup problem, and which number to use for each decision.
The Main Causes
Self-attribution. Self-attributing networks such as Meta, Google and Apple Ads count every install where their ad was involved inside their own window. They do not know when another network's click came later. Your MMP sees every network and credits one source per install, usually the last eligible click. Add up every network's self-reported installs and the total will exceed your real install count.
Window differences. If your MMP uses a 30-day click lookback window and a network uses 7 days, an install 10 days after the click counts in the MMP and not in the network. The reverse happens when the network's window is longer.
View-through and engaged views. Networks often include installs after an impression or a video view with no click. An MMP may not credit those, or may credit them only where view-through is turned on.
Cross-device. Logged-in platforms can credit a click on one device for an install on another. An MMP matching signals on the same device will not.
Modeled conversions. On iOS, networks often add modeled and aggregated conversions to fill gaps left by ATT and SKAN. MMP device-level attribution is limited to what it can match.
Date basis and time zones. Networks often report conversions on the click or impression date; MMPs often report on the install date. A one-day shift in time zone or date basis moves numbers between days.
Event and time limits. Networks may reject events sent too long after they happened. Those events stay in the MMP but never reach the network.
How to Reconcile
- Compare like with like. Installs to installs, by the same date basis, in the same time zone.
- Align windows where you can, or note where they differ.
- Split by type. Separate click, view-through, engaged-view and modeled conversions in network reports.
- Check by platform. Android gaps are usually smaller than iOS gaps.
- Track the gap over time. A steady gap is a known quantity. A sudden jump points to a broken integration, missing credentials or a tracking change.
Which Number to Trust
Use each number for what it is good at. Network-reported conversions are what the network's bidding learns from, so they matter for in-platform optimization. MMP numbers credit each install once across every channel, so they are the right basis for comparing channels, calculating blended CPI and ROAS, and deciding where budget goes.
Discrepancies in Linkrunner
Linkrunner credits each install to one source with last-touch rules, deterministic matching first and IP and user-agent fallback only when no deterministic signal exists. Click windows default to 30 days for Google and Meta and 7 days for other networks, and can be set from 1 to 30 days per channel, which helps align with network settings.
The docs include discrepancy guides for Meta and Google that walk through the expected causes, such as Meta's default 7-day click and 1-day view settings, view-through on Android only when Meta Install Referrer is set up, same-device matching, and Meta's 7-day limit for accepting events. Postback delivery logs show what was sent to each network and when. See mobile attribution and Google and Meta ads.