Skip to main content
Attribution is the practice of giving credit for a purchase to the marketing activity that led to it. You can attribute a purchase to any channel, campaign or ad you run. Attribution shows which activities produce a return, so you can decide where to invest next. An attribution model is the set of rules that decides how credit for a purchase gets split across the measured touchpoints. Models are often expressed in monetary value. If a purchase is worth EUR 100, the model spreads that value across the measured touchpoints according to its own rules. The right model depends on your business model, your strategy and the questions you are trying to answer.

​
First touch and last touch

First touch attribution gives all credit for a purchase to the first touchpoint measured. Last touch attribution gives all credit to the last touchpoint before the purchase. Neither model splits credit across touchpoints, but both show which channels create initial awareness and which channels close the sale.

​
Linear

Linear attribution spreads the credit for a purchase evenly across all measured touchpoints. If a customer clicked five ads before buying, each ad gets 20% of the credit. Linear attribution shows your full marketing mix and which touchpoints move customers through the funnel toward a purchase.

​
Time decay

Time decay attribution gives more credit to touchpoints closer in time to the purchase. It gives less credit to touchpoints further away. This model assumes that touchpoints later in the funnel have more impact on the purchase, while still crediting earlier touchpoints. Time decay helps you understand and optimize the different phases of the customer journey.

​
U-shaped

U-shaped attribution gives most of the credit to the first and last touchpoints measured before a purchase. It spreads the remaining credit evenly across the touchpoints in between. Use this model when you have a clear picture of your typical customer journey. It optimizes for both top-of-funnel input and bottom-of-funnel conversion.