Recommendations
The ExpertSender Web Module recommendation system allows you to always display the most relevant product recommendations to your customers.
As the setup of the recommendation engine is carried out by ExpertSender, this article provides only general information on how the functionality works.
The recommendation engine monitors the website traffic patterns and actions of the visitors, then feeds this data to the recommendation model(s) connected to a given website. The recommendation engine usually needs two weeks to teach and train each model about its users.
A single website can rely on multiple recommendation models. Depending on their settings, they’ll use different algorithms for generating recommendations. For example, one model may perform better when its recommendations are used in abandoned cart events, while a different one will generate better results when used in a popup campaign.
Each model can be used to generate recommendations for:
- Events sent to Brand subscribers
- Events sent to Reachable non-subscribers (retargeting must be enabled)
- Popup campaigns (popups must be enabled)
- Exports to Data Tables in ExpertSender
Recommendations added to events and pop-ups are called live recommendations, as they are generated at the moment an event is sent or a popup is displayed. They use the most up-to-date data available. In contrast, recommendation exports to Data Tables are done periodically, which means that it can sometimes happen that, for example, recommendations in the table will suggest the same product as the one that a subscriber bought just an hour ago.
Please note that the full potential of the recommendation feature can be leveraged only if there is heavy traffic on your website (at least 200,000 visitors per month) and if your product palette is sufficiently broad. Otherwise, there may not be enough data for the recommendation algorithms.