Calibrated Attribution

The daily reading, campaign by campaign. The MMM coefficient calibrates the platform numbers.

Last click
Calibrated

The problem with traditional digital attribution

Classic attribution (last click) gives the whole conversion to the last touchpoint. Branded search and retargeting are over-credited. Awareness, TV and offline are not seen.

Budget then moves to retargeting, and is missing from the drivers that create demand.

How does Calibrated Attribution work?

Calibrated attribution applies the MMM coefficient to platform data, campaign by campaign, every day.

An example. Google reports a ROAS of 10, 12 and 8 for your three Google campaigns. The MMM, which measures Google’s contribution alongside all your other drivers, online and offline, puts Google’s overall ROAS at 6. Calibrated attribution applies that coefficient to each campaign, so the total comes back to 6. The ranking between campaigns is kept: your best campaign stays the best, at its true value.

CampaignROAS reported by GoogleCalibrated ROAS
Campaign A106
Campaign B127.2
Campaign C84.8
Google, total106
Illustrative example, with equal spend across campaigns.

Your attribution data, calibrated

We don’t produce attribution data. We start from the data your teams already use: attribution reported by your platforms (Google, Meta, TikTok, Amazon…), your analytics tool (Google Analytics, Adobe, Piano Analytics) or an in-house tool.

Last click, post-view or multi-touch: the attribution method matters little, as long as it is the same across every channel you compare.

Why the correction usually goes down

Each platform measures what it sees. When a customer sees an ad on Meta, then clicks a Google ad before buying, both platforms claim the same sale. Added together, the conversions your platforms report exceed your actual sales.

The MMM allocates 100% of sales: the baseline, the incremental contribution of each media channel and that of the other variables. Each platform, on the other hand, claims sales on its own side: added up, their figures can exceed 500% of actual revenue.

The MMM starts from your actual sales and brings the total back to what really happened. The reverse happens too: some channels are under-credited by their own platform, and their calibrated figure goes up.

Calibrated cross-platform attribution

mediaROI’s calibrated attribution is cross-platform: it covers your digital channels, including Google, Meta, TikTok, and Amazon, and aligns them with the reality measured by MMM. You get a unified, reliable view of campaign performance every day.

Daily

Attribution data updated every day, not once per quarter.

Cross-platform

Reliable comparison across Google, Meta, TikTok, Amazon and all your channels.

Bias-corrected

Reduce retargeting over-crediting and restore a fairer read on lower-funnel levers.

What you see in the platform

For each channel, the Attribution page shows:

  • spend;
  • the contribution as reported by the platform;
  • the contribution calibrated by the MMM;
  • the correction coefficient applied.

The table can be exported for your own analyses or reporting tools.

One prerequisite

An MMM run by mediaROI and updated every month

It provides monthly coefficients. Coefficients from a model frozen several months ago would produce wrong numbers. The attribution data itself comes in through the platform connectors.

Calibrated attribution is included in the monthly MMM update plan, at no extra cost.

Where each tool stops

The MMM gives the contribution of each driver and each platform. It does not go down to campaign, audience or creative level. Calibrated attribution is the daily tactical reading: it goes down to campaign, creative and day, and keeps the MMM figures as the reference. You keep the granularity: campaign by campaign, day by day.

It does not replace a specialist attribution tool for fine optimization within a channel: bids, audiences, formats. It is not meant to split your annual budget; that is the MMM’s job. And it does not prove a causal effect; that is what an incrementality test is for.

Being explicit about the boundary is part of the method: it is what lets you trust the number.

FAQ

Frequently asked questions

Does calibrated attribution replace my attribution tool?
No. The MMM does not replace your attribution tool, it makes it reliable. Your attribution data remains the working base; the MMM corrects its biases.
Do I need to change my attribution method?
No. Last click, post-view or multi-touch: we start from the method you use, as long as it is the same across every channel you compare.
How often are the figures calibrated?
Your attribution data is calibrated every day. The correction coefficients are updated with each monthly MMM refresh.

Discover the platform

30 minutes to walk through a complete model and ask your questions