What is multi-touch attribution (MTA)?
How does it work?
Attribution means assigning credit for a key action, such as a purchase, to the ads, clicks and other touchpoints on the path before it. "Multi-touch" means more than one of those touchpoints gets a share.
The split depends on the model's rules, and different rules give different answers for the same sale.
Why does it matter?
An agency's measurement guide treats attribution and incrementality testing as separate methods: attribution is fast and granular, suited to day-to-day optimisation, while a test isolates cause1. That is the point. MTA tells you how credit is divided among the touchpoints it can see. It cannot tell you whether the sale needed any of them.
That limit has practical consequences:
- Ad views. A model may count only clicks, or it may give views a share. Both are choices, and they change the answer.
- People who decline tracking. A person who declines tracking leaves no path to split. The model sees only the part of your customers it can follow.
- Sales the platforms also claim. Two tools can credit the same sale to different ads. Added together, the credit can exceed the sales.
Our view: use MTA for what it is good at, which is a quick, detailed look at paths. Use a test when the question is whether the ads caused the sale.
Worked example
Easily confused with
Multi-touch attribution vs Data-driven attribution (DDA)
Google's attribution model that uses machine learning on converting and non-converting paths to estimate how each interaction affects the probability of conversion. It is built separately for each advertiser and key event. One model, not the whole category.
Multi-touch attribution vs Marketing-mix modelling (MMM)
A statistical method that estimates how marketing inputs contribute to results using aggregated data, without user-level tracking. MTA works on individual paths. MMM does not.
Multi-touch attribution vs Incrementality
The extra sales caused by a marketing activity, over and above what would have happened without it. MTA does not answer that question.
How these three compare is covered in MMM vs MTA vs incrementality tests.
FAQ
Is multi-touch attribution the same as incrementality?
No. MTA divides credit among touchpoints. Incrementality asks what the ads added beyond what would have happened anyway.
Is multi-touch attribution more accurate than last click?
It shares credit across more touchpoints, so it is less blunt. Less blunt is not the same as correct. Both rules assign credit without testing cause.
Is multi-touch attribution the same as multi-channel attribution?
The two names are often used for the same idea. We use "multi-touch attribution", because the credit is split between touchpoints, not just channels.
Can MTA replace a test?
No. It can show where to look. A test is how you check the answer.
Let’s find what’s actually working.
Next step. Whatever the attribution model says, check what the result is worth after costs. Use the Is your ROAS real? calculator.