CHATTERgo recommendations put the right products in front of each shopper — in the chat and on your storefront shelves — and measure what they earn.
How strategies pick products
Every recommendation row is driven by a strategy, and each strategy family picks products differently:
- Collaborative filtering (CF) — learns from shopper behaviour across your store: Bought together, Visitors also viewed, From your recent views, and Recommended for you (built per shopper nightly from their own interactions).
- Popularity — behavioural aggregation such as Trending now.
- LLM-driven — picks from the shopper's profile and conversation, like persona-based suggestions and personal search queries.
- Rules — deterministic catalog and context matching: same brand, same collection, preferred brand or collection, accessory gaps, repurchase reminders, cart add-ons, and a discovery fallback.
Collaborative filtering needs interaction data to learn from. New stores start on popularity and rule strategies; as conversations and views accumulate, the nightly per-store model build begins serving CF strategies automatically.
Choosing strategies per placement
The experience matrix in your dashboard lists your placements (product pages, home page, cart, and so on) and lets you add or remove strategies per placement. A balanced default works out of the box; the advanced editor is reserved for admin and owner roles.
Measuring results
The analytics pages report impressions, clicks, add-to-cart, purchases, and revenue per row and per strategy — see Analytics overview. Use those numbers when deciding which strategies to keep on a placement.