Attribution Models

Attribution models are the rule sets used to distribute conversion credit across multiple marketing touchpoints in a customer's journey.

Last-touch gives 100% credit to the final touchpoint before conversion — simple, but it makes upper-funnel channels (awareness, social, content) look worthless even when they're doing real work bringing people into the funnel. First-touch does the reverse, over-crediting discovery channels and under-crediting the channels that actually close the sale.

Linear splits credit evenly across every touchpoint; position-based (U-shaped) weights the first and last touch more heavily than the middle ones. Data-driven attribution (DDA) — GA4's default model — uses machine learning trained on your own conversion paths to algorithmically weight each touchpoint's actual contribution, generally considered the most accurate available option, though it requires meaningful conversion volume (Google states roughly 300+ conversions and clicks across at least two ad channels within 30 days) to work reliably.

No model, including data-driven, escapes the fundamental limitation described in attribution — they all measure correlation with observed touchpoints, not proven causation.