Measurement Architecture • July 2, 2026 • 8 min read

Why Ad Network and MMP Install Reports Never Match: The Mechanics of Attribution Discrepancies

Deconstructing Self-Attributing Networks, lookback window drift, and time zone mismatching.

The Common Attribution Conflict

Every mobile growth leader has encountered the spreadsheet where Google Ads claims 5,000 installs, Meta Ads claims 4,200 installs, Apple Search Ads claims 2,100 installs, and yet the Mobile Measurement Partner (MMP) reports only 8,000 total paid installs combined.

Where do the missing 3,300 installs go? Are ad networks fabricating conversions, or is the MMP dropping data?

The answer lies in the structural divergence between self-attribution modeling and last-touch cross-channel deduplication.

1. The Self-Attributing Network (SAN) Walled Garden

Major ad channels—including Google App Campaigns, Meta Ads, Apple Search Ads, and TikTok—operate as Self-Attributing Networks (SANs). Rather than utilizing third-party tracking links with redirect parameters, SANs receive raw install signals from your MMP via server-to-server postback APIs.

Upon receiving an install event, the SAN queries its internal database for all user ad interactions (impressions and clicks) that occurred within its configured attribution lookback window. If a match is found, the SAN responds to the MMP: “We claim this install.”

Critically, the SAN evaluates this interaction in total isolation from other media channels. If a user clicked a Google ad on Monday and an Apple Search Ad on Tuesday before installing, both networks claim 100% credit in their native reporting interfaces.

2. The MMP’s Role: Cross-Channel Arbitration

The MMP acts as an independent referee. Applying a single last-touch attribution rule, the MMP evaluates the timestamps of all incoming claims:

[Monday 10:00 AM] User clicks Google Ad (Claim registered)
[Tuesday 04:00 PM] User clicks Apple Search Ad (Claim registered)
[Tuesday 04:15 PM] User opens App (Install fired)

Result:
- Google Console reports: 1 Install
- Apple Search Ads Console reports: 1 Install
- MMP reports: 1 Install attributed to Apple Search Ads, 0 to Google

In this standard scenario, Google will bill the advertiser for the conversion and display the install in Google Ads Manager, while the MMP attributes the conversion to Apple Search Ads.

3. Lookback Window Inconsistencies

Discrepancies escalate when attribution lookback windows are misaligned between platforms:

Click-Through Windows: A network may default to a 30-day click window, while the MMP is configured to credit clicks only within a 7-day lookback.
View-Through Attribution (VTA): Ad networks often default to 24-hour impression-based view-through attribution. If a user scrolls past a video ad in their feed and subsequently downloads the app hours later via organic search, the ad network claims an install even though no deliberate engagement occurred.

4. Time Zone and Cohort Date Definition Mismatches

A third source of reporting friction is temporal misalignment:

Interaction Date vs. Install Date: Ad networks frequently attribute installs back to the date and time of the ad click (interaction time). In contrast, MMPs attribute installs and downstream cohort revenue to the exact timestamp of the initial app open (install time).
Time Zone Offsets: Comparing UTC reporting in your MMP against PST reporting in an ad console creates rolling date-boundary discrepancies of up to 30%.

How to Establish Attribution Clarity

Standardize Lookback Windows: Enforce uniform click-through (7-day or 1-day) and view-through (disabled or restricted to 12-hour) windows across all integrated partners within your MMP.
Audit Raw Postback Logs: Inspect raw device-level touchpoint logs to quantify the exact overlap percentage between SANs and independent ad networks.
Calibrate with Incrementality: Run controlled geo-holdout tests on channels showing disproportionate VTA claims to determine whether impressions genuinely generated net-new installs.