When you enable a Best next recommendation, the model analyses your order data to identify patterns in customer purchases. It builds a matrix of customers and products, grouping together those with similar buying habits. Then, it uses this data to uncover products that one customer has bought but another has not, highlighting the items that may be of interest to those customers.

This approach works well for retailers with limited product data, as it relies on order history rather than product attributes. Best next is quick to set up and requires minimal data to generate recommendations.

If your account does not have predictive recommendations enabled:

A side panel allows you to request that the feature be enabled on your account (for an additional monthly charge). We review your data first, then confirm whether the feature is ready to use or if there are data dependencies to resolve.

Predictive recommendations are available on a 30-day free trial.

Learn more in review the data dependencies for this product recommendation type.

This model works well even with product catalogs that have limited detail. It only considers order data for making recommendations. The more order data you have, the higher the accuracy of the recommendations.

Best next requires less data than the Lookalikes recommendation model, which uses content-based filtering, so it generates recommendations quickly. It provides good quality recommendations for most types of retailer.