Item recommendations can be implemented into different parts of your system to provide your customers with an omni-channel personalized experience and increase the likelihood of repeat purchases. Our recommendation models are trained using our proprietary hybrid recommendation system. Hybrid recommendation systems are the most powerful recommendation systems currently available. Items are recommended based on a profile's previous interactions with items as well as the item interactions of similar profiles.
Recommendation models are trained and updated at regular short-term intervals to ensure the items recommended are always the most relevant to each profile.
Item recommendations are constructed by making a request to the /v1/retention/recommendations ↗️ endpoint.
If you wish to predict item recommendations, an array of
profile_id(s) MUST be supplied within the body of the request.
Up to 10* item recommendations can be made per profile identifier for a maximum of 100 profile identifiers with any single request to the /v1/retention/recommendations ↗️ endpoint.
*This is dependent on the number of active items associated with your account.