Marketing mix model results and platform attribution numbers almost always disagree. The gap is expected and not a sign that either is broken. Understanding why they differ is what allows you to make good decisions from both data sources.
A common experience: your MMM says paid social contributes 15 percent of revenue. Meta Ads Manager says paid social drove 35 percent of attributed revenue. Which is right? Both are right, within their own frameworks. The question is which framework is appropriate for the decision you are making.
Why the numbers differ
Platform attribution systems count every conversion associated with a touchpoint and credit it fully or fractionally to that platform's ads. They are designed to show the value of advertising on that platform, using that platform's data.
Marketing mix models estimate the incremental contribution of each channel to total revenue using statistical analysis of historical data. They are designed to separate the causal effect of each channel from baseline demand and seasonality.
The three main sources of disagreement
- Attribution overlap: multiple platforms claim credit for the same conversion because each touched the user at some point in the journey
- Baseline revenue: MMMs estimate a baseline level of sales that would occur without any advertising. Platform attribution does not subtract this baseline, so it counts conversions that would have happened anyway
- Brand halo effects: MMMs can separate organic brand demand from paid advertising effects. Platform attribution counts organic search purchases as paid search conversions when a branded paid search ad was present
Which source to trust for which decisions
Platform attribution data is more useful for day-to-day operational decisions: which creatives to scale, which audiences to target, and how to allocate budget within a platform. It is a fast, granular signal that is well-suited to tactical optimisation.
MMM results are more useful for strategic budget allocation decisions: how much to spend across channels, how to rebalance the media mix, and how to justify marketing investment to finance. MMM provides a cross-channel, deduplication-adjusted view that platform data cannot.
When your MMM shows a channel contributing far less than its platform-reported attribution, that is useful information, not a problem to solve. It usually means the channel is claiming credit for demand it did not create. The MMM is separating the causal contribution from the coincidental association.
Practical reconciliation steps
Start by comparing the sum of all platform-attributed revenue to your actual total revenue for the same period. Calculate the overclaim factor. A factor of 1.3 means platform attribution is claiming 30 percent more than your actual revenue. This tells you how much inflation to expect when reading platform data.
Next, compare the relative contribution of each channel in the MMM to its share of attributed revenue in platform data. Channels where the MMM share is significantly lower than the platform attribution share are the ones most likely over-claiming. Channels where the MMM share is higher than the platform attribution share may be under-credited in your current setup.
Using incrementality to validate both
When the MMM and platform data disagree substantially on a high-spend channel, an incrementality test is the best way to determine which is closer to the truth. A geo holdout test produces an experimental lift estimate that can be compared to both the MMM coefficient and the platform-attributed ROAS.
The experimental result is your most credible reference point. Adjust your reliance on both the MMM and the platform data in proportion to how well each aligns with the experimental finding.
Is it normal for MMM to show much lower channel contributions than platform attribution?
Yes, consistently. MMMs typically show lower contributions for digital channels, particularly paid social and display, compared to platform-attributed numbers. This is because MMMs subtract baseline demand and avoid double-counting across channels. The gap is often 30 to 60 percent for upper-funnel digital channels.
How do I explain the gap to my paid social agency?
Frame it as a different measurement question. Platform attribution answers: "which conversions were associated with a paid social touchpoint?" MMM answers: "how many additional conversions did paid social cause beyond what would have happened without it?" Both numbers are real. They measure different things. Budget decisions should be based on the causal question, which is what the MMM addresses.
Should I share MMM results with my media platforms?
Sharing MMM results with platforms can improve their optimisation if they accept external signals. Meta's Conversions API and Google's offline conversion imports allow you to feed first-party data back to platforms. Sharing the fact that your MMM shows a channel at a different efficiency level than its own reporting is a useful conversation to have with your agency, even if the platform algorithm cannot directly use the MMM output.