The recommendation engine integrates information from the DMP, regarding the products visited by the user, and matches these products to the partner’s product catalogue, imported into our Feed Crawler. By cross-referencing the information that the product seen by a certain user is present in the imported catalogue, the recommendation engine ensures that the product is displayed later in a banner to that user.
More than that, our recommendation engine is optimized to deliver conversions, not just clicks. To do so, factors such as last visited products, average ticket, category, most desired and most purchased products, including also complementary ones to those seen or purchased, are taken into consideration. In addition, we estimate the average ticket to make a recommendation based on a 30, 60, and up to 360-day history.