Understanding the affinity score

Learn what we mean by affinity score and what goes on behind the scenes to calculate it.


The affinity score is a percentile measure of how confident we are that a particular product fits the wants and needs of an individual customer. You can see the affinity score when you preview a product recommendation against a particular customer.

To learn how to preview a product recommendation, check out the article Create a product recommendation.

How we calculate the affinity score

We calculate the affinity score in different ways, depending on which recommendation you are using. But in general, we detect trends in product characteristics and purchasing and use these trends to determine which products to recommend to which customers.

Example: Best next

For the Best next recommendation, the affinity score shows how well the products fit the predicted customer journey. We calculate this affinity score based on which, and how many, similar shoppers bought these predicted products in past purchases – this is called collaborative filtering.

Example: Lookalike

For the Lookalike recommendation, the affinity score shows how similar products are to the customer's previous purchases. It looks at a product's characteristics, such as name, brand, colour, and type.

See also

Did you find this article helpful?

Can we help?

Thanks for using Dotdigital. If you need more help or support, then contact our support team.