Attribution model
Also known as: marketing attribution, attribution modeling
An attribution model is the rule a company uses to decide how much credit for a sale or a lead goes to each marketing touchpoint the customer interacted with before converting.
A customer rarely converts after seeing a single ad. They might click a paid search ad, read an email three weeks later, and finally convert after a direct visit to the site. An attribution model decides which of those three touchpoints gets credit for the sale, and different models give a very different answer.
Linear model: credit per touchpoint = 1 / number of touchpoints in the journeyExample
Take a customer with three touchpoints before converting: a paid search click, an email open and a direct visit. Under a last-click model, the direct visit gets 100% of the credit and the other two get none. Under a linear model, each of the three touchpoints gets one third of the credit, or about 33%. If the company only ever looks at last-click, paid search appears to contribute nothing to this sale, even though it started the journey.
The split shown here illustrates how the two models diverge on the same journey, not a measured result from a real campaign.
How to use it
Pick the model based on the decision it needs to support, not out of habit. Last-click is simple and fine for a business with a short, single-channel journey, while linear or a position-based model fits a longer journey with several channels involved before a sale. Whatever model is chosen, use the same one across every report so channel comparisons stay consistent month to month.
Re-check the model choice when the buying journey changes length, since a model built for a two-touch journey can badly misrepresent credit once the journey grows to six or seven touches.
Common mistakes
A common mistake is switching attribution models between reports to make a favored channel look better, which destroys trust in the numbers once someone notices. Another is relying only on last-click attribution for a business with a long consideration period, which starves early-funnel channels like content or awareness campaigns of any credit and budget. A third is treating the model’s output as fact rather than as an estimate, since no attribution model perfectly captures what actually influenced a buyer’s decision.
Pair attribution with CAC payback period by channel before cutting a channel that looks weak under one model but strong under another.