Here's all you need to do to integrate our recommendations platform into your tech stack to create engaging experiences at every customer touchpoints.

‍

Item Catalog

Enriching the information our algorithm can have about your items is strongly recommended, although also optional!

Using rich properties for your items offers two advantages:
1 - It improves the recommendations, especially for both cold-start problems where the algorithm relies only on properties (such as Semantic Graph Embedding from genres and tag, or Deep Content Extraction from text and images)
2 - It enables you to dynamically filter the recommendations on items satisfying certain criteria (such as a price smaller than a threshold given at runtime)

User Data (optional)

Naturally, providing information about your consumer (while respecting privacy and security) can help the recommendation engines build your consumers' DNA.
‍
Sending user data is OPTIONAL but can improve the recommendation quality, especially when it comes to cold-start recommendations where your user hasn't interacted with your products yet.

‍

Behavior Data

The only data REQUIRED to train the best recommendation engines for your business is, of course, the interactions the users have with your products. You can, of course, upload those interactions in 2 different way:
‍
Direct User Interactions: or what we call "Implicit Feedback." Those are all the interactions you can find, for instance, in your Google Analytics, the clicks, the scroll, the page view, etc. User Interactions represent different interactions a user may have with an item, often hints whether the user likes or not an item.

‍

Ratings or Explicit Feedbacks: Explicit feedbacks clearly describes on a fixed scale the rating a user gave to an item (like/dislike, star rating, etc.). For more information about how to prepare your file and preprocess them.
‍