Data-Driven Attribution

Data-driven attribution is GA4's default model, which uses machine learning to distribute conversion credit across touchpoints based on their measured contribution.

Instead of a fixed rule, data-driven attribution (DDA) looks at your own conversion paths and compares paths that converted against paths that did not. Touchpoints that move the needle get more credit, the rest get less. It is the default for every GA4 property, and unlike the old rules-based models, the split is not something you can predict by hand.

Two paths both end in a purchase. One went Paid Search then Direct, the other went straight to Direct. If Direct converts fine on its own but Paid Search consistently precedes purchases that would not otherwise happen, DDA leans credit toward Paid Search. You do not set those weights. The model derives them from your account's data.

DDA needs volume to work, and it is a black box. Google's model requires meaningful conversion data before it produces stable results, so thin accounts get noisy, shifting credit. And because you cannot see the weights, you cannot fully audit why a channel's credit moved. Do not treat month-to-month swings in DDA as gospel when volume is low.

Default does not mean interrogation-proof.