Use case · Recommendations
The second item in the basket.
AACSearch learns from visits which products are viewed together, opened next and bought together, and serves them as recommendation rows — diversified, explainable and filtered by what may not be suggested.
Rows that work
Frequently bought together
Items added to the cart or bought in the same visit, weighted so a bestseller doesn't pair with everything.
Similar items
Close alternatives by embedding — the answer when the chosen size or model is out of stock.
For returning visitors
Ranking reshaped by each visitor's decayed history of clicks and queries, with consent respected.
Control
Never the wrong accessory
Manual excludes links remove incompatible items from every row that starts from that product.
Measure it
Split traffic between two profiles and compare clicks and conversions before rolling a change out.
Compatibility from the catalogue
Will derive requires and excludes from catalogue attributes instead of manual links.
Specifications
- Bought-together evidence
- 2 or more visits with both items in the basket
- Fallback
- similar items until there is enough data
- Incompatible items
- removed by manual
excludeslinks
Before you start
- Add-to-cart events from your backend
- A product page or cart slot for the recommendation row
How to start
- 1
Send events
The widget records clicks and views; send add-to-cart and purchases from your backend.
POST /api/v1/indexes/products/events { "type": "add_to_cart", "document_id": "sku-123", "session_id": "visitor-42" } - 2
Ask for a row
Bought-together for the cart, similar for the product page.
POST /api/v1/recommendations { "index": "products", "seed_document_id": "sku-123", "strategy": "complement", "limit": 6 }