What Is Attribution? (Plain English, No Jargon)

The short version

• —A customer usually touches several channels before buying. Attribution decides how to split the credit between them.
• —Last-click attribution — still the most common default — gives 100% of the credit to whatever they clicked last.
• —That systematically over-rewards the bottom of the funnel (brand search, retargeting) and under-rewards whatever created the demand.
• —Every ad platform runs its own attribution, and each one credits itself. That is why the numbers never reconcile.
• —Getting credit right matters because it decides where next quarter's budget goes.

Why this happens

Attribution is hard because you can only observe one version of history. You can see that someone saw an ad and then bought. You cannot see whether they would have bought without it. Every attribution model is an attempt to guess that missing half, and the simple models guess badly — they use a rule of thumb (last click, first click, evenly split) instead of measuring anything.

The proper name for the missing half is the counterfactual: what would have happened otherwise. Methods that estimate it — holdout tests, geo experiments, causal inference — answer a fundamentally different question from rule-based attribution. Rules ask "who was there when the sale happened?" Causal methods ask "which spend actually changed the outcome?"

If you have found this confusing, that is a reasonable response to a genuinely confusing situation rather than a sign you are missing something obvious. The industry uses one word, "attribution", for both a bookkeeping rule and a scientific question, and most dashboards never tell you which one you are looking at.

What you can do about it today

1. 1

   Find out which model your reports use

   In GA4, check the attribution settings; in Meta and Google, check the attribution window. You may find the tools you compare every morning are not using the same rules, which alone explains a lot of the disagreement.

2. 2

   Get your UTMs consistent

   Attribution of any kind is only as good as the labels on your traffic. One convention, applied everywhere, lowercase, no variations. This is unglamorous and it is the highest-leverage hour you will spend.

3. 3

   Compare a channel's claim against your total

   Sum what every platform claims and compare it to your real order count. Seeing the overlap once makes the whole problem concrete in a way no article can.

How Causality Engine helps

Causality Engine skips the rules and estimates the counterfactual directly from your GA4 export, so you get a per-channel view of what actually caused revenue rather than which channel was standing closest to the sale. €99 per read, 5-10 minutes, no pixel to install.

Questions people ask next