Skip to content

What is attribution model?

An attribution model is the rule your reporting uses to divide credit for an order between the different marketing touchpoints a customer encountered on the way to buying.

By cartscaling Reviewed

Worked example, not a result

One order, three touchpoints, three answers

Say a customer sees a social ad, comes back a few days later through a search ad, and finally arrives through a link in an email before placing an order worth 228.

Under a last click rule, the email takes the full 228 and the two ads take nothing. Under a first click rule, the social ad takes all of it. Under an even split, each of the three touchpoints takes 76.

Nothing about the customer or the order changed. Only the accounting rule changed, and each version would lead a store to a different budget decision. Last click makes email look extraordinary and prospecting look wasteful, which is how stores end up cutting the campaigns that introduce them to customers in the first place.

That is why the model has to be chosen deliberately and then held still, and why a total business measure such as revenue against total marketing spend is a useful check on whatever the models say.

The order value here is an example.

Why it matters

Attribution decides where budget goes, so it quietly decides how the visitors lever is managed. A model that overweights the last touch will always favour retargeting, brand search and email, because those sit closest to the purchase, and will always underfund the campaigns that create demand in the first place.

It also explains most disagreements between platforms. Each ad platform reports on the conversions it had a hand in, using its own window and its own rule, so the totals overlap and the sums add up to more revenue than the store actually took. That is not dishonesty, it is three different accounting systems describing the same order.

For a store owner the practical position is to pick one model as the internal standard, use platform reports for direction within a channel rather than for cross channel comparison, and settle real budget arguments with total revenue against total spend.

Common mistakes

  • Adding conversions from several platforms together. The same order appears in more than one report, and the combined figure describes a business bigger than yours.
  • Changing the model and comparing across the change. The reported performance of every channel moves, and it is easy to read an accounting change as a real one.
  • Using last click and then cutting whatever it shows as weak. That reliably removes the campaigns that introduce new customers, and the effect shows up a month or two later.
  • Trusting long lookback windows without thinking. A window longer than your buying cycle credits ads for orders that were going to happen anyway.

Related terms and answers

Your plan is free.

A person reads it. You keep the plan. No logins, no call first.

Store, name, email and phone. A person replies from a real inbox. Privacy

Get your free growth plan