AI recommenders (and personalization in general) are used to suggest pertinent content to your website’s visitors. They work by being trained using data that is passively being sent by visitors (called user events) so the AI can learn about their behaviours and needs. Some of the most popular recommenders are Algolia, Recombee and Google Retail.
They are a great way of increasing CTR. Here are some of the 7 best events to train your AI recommenders :
Detailed view of an item (clicked)
This user event is sent whenever a user views any page or product on your website.
They are valuable for recommenders since they can give insights into what the user might be interested by and is one of the first steps of the conversion funnel.
This type of recommendation can be used to solve the problem of cold-start (with collaborative filtering) since page view events are plentiful (since one is sent for every page visit). This means that even new items can quickly gather data be recommended to the correct visitors fast.
Detailed view of an item from a recommendation (clicked)
The ‘detailed view of an item from a recommendation‘ user event is sent whenever a user clicks on an item from a recommendation. This is considered a high value event and is deemed a success, as the recommender suggested the appropriate content for this particular visitor. It provides incredible feedback since the recommender can gauge its effectiveness and course correctness.
Category page view
The ‘category page view‘ event is triggered whenever a user views a page designed to display items from a specific category. This is so the recommender can tell if specific categories are of interest to some users, which is more precise than the ‘detailed view’ event because the reason why a user clicked on a specific item can be for multitude of factors. Whereas if a user visits a category page, there is only one reason : because they are interested in this category.
For example, it could be a page of category ‘Car’ where only the items of category ‘Car’ are displayed.
Add to favourites/Bookmark
The ‘add to favourites‘ event is sent whenever a user adds to favourites or bookmarks content. This means that this item is interesting enough and the user wishes to find it in the future. This reflects a strong level of interest since the user values the item enough to want to save and review it later. You can then use these events to identify trends. This type of user event is considered a success or a positive for the recommender.
Add item to cart
The ‘add to cart‘ event is a necessity for e-commerce websites. It will be sent whenever the customer adds an item to cart, and is very useful for recommenders. Since it is close to the end of the conversion funnel, this is an item that is highly convincing and relevant to a specific category of users.
Purchase an item
‘Purchased an item‘ event is the most important event for e-commerce websites since it means it is that a conversion has been reached. Once again, this event triggering means that the recommended product is extremely relevant so this event is the most valuable there is.
Add rating
‘Added rating‘ is sent whenever a user gives a rating. For example, the user gave 4 stars to a product. As you can guess, this rating/score can then be used by the recommender to determine if similar users/customers would want to view this content.
