About Session-Based Product Recommendations (SBPR)

This is for:

In this article

Prerequisites

Coveo Machine Learning (Coveo ML) Session-Based Product Recommendation (SBPR) models suggest the products that are most likely to be relevant given the user’s current session behavior on a Coveo-powered commerce interface.

Traditional product recommendation models learn product relationships from historical interaction patterns across users and sessions. SBPR models enrich these established behavioral signals with the user’s current shopping session and commerce interactions, allowing recommendations to better reflect the shopper’s in-the-moment intent.

SBPR models process the user’s recent commerce interactions to update the recommendations provided in real time. The model continuously processes the actions taken by the user to recommend products that are relevant to their current shopping session. A cold-start user, including one who isn’t authenticated, can still see relevant and personalized recommendations, because the model infers their intent as soon as they start interacting with the interface.

Note

A model can take anywhere from several minutes to several hours to build, depending on the size of the model dataset and the number of model build requests in the queue (system load).

Example

During a shopping session on a Coveo-powered sporting goods commerce interface, an unauthenticated user performs the following actions:

Enters Golf iron sets in the search box.
Navigates through the iron sets that are being displayed on the results page, and then clicks on mid-range golf iron sets.
Enters Golf driver in the search box, navigates through the available drivers, and then returns to the home page.

By analyzing the user’s session, the model understands that they’re currently looking for mid-range golf equipment. Therefore, since the recommendations interface on the home page leverages an SBPR model, the products displayed to the current user are complementary products, such as mid-range golf gloves and balls.

Prerequisites

Contact your Coveo representative to enable SBPR in your Coveo organization.

To use SBPR, you must meet the following prerequisites:

Your organization must contain at least one Catalog source.

Note

The field that uniquely identifies the items in the Catalog source, such as ec_product_id, must be mapped to the permanentid field.

To train Coveo Machine Learning (Coveo ML) models, you can’t use the # character in either of these fields. See Manage source mappings for instructions on how to define mappings on a source.

Your commerce interface must track, at a minimum, the following commerce events: