The problem
A checkout page has one chance to raise the order value, and it has to decide from the whole basket, not from one product. Product-to-product "bought together" lists answer a narrower question; the basket as a whole is the better signal of what the shopper is missing.
How it works
The basket's categories go into the where clause and _relate over past orders returns the categories that co-occur with that combination, ranked by lift. The application drops weak relations, picks a product from each strong category, and shows the expected uplift as the attach probability times the price.
It is the same engine as Bought Together, asked about the whole basket instead of a single anchor product.
For the full architecture, see the technology overview. For the broader narrative across multiple use cases, read The Predictive Application.
See it live
This use case runs in the 🛒 E-commerce demo today. Click through to the live application and inspect the queries that produce the result.
