×![]()
Click anywhere to close
Recommendations learn from three events: product viewed, added to cart, and purchased. Views and carts drive the ranking; purchase does two jobs — it attributes revenue to your campaigns and completes the picture of what actually converts. (Content sites track article viewed the same way.) This page covers every way to wire these events, from zero-code to API.
Pick your route
| Route | Best for | Setup |
|---|---|---|
| Platform integration | Shopify / WooCommerce stores | One toggle: Shopify events (and the cart extension), or the WordPress plugin’s add-to-cart and purchase toggles. |
| GTM dataLayer | Sites already reporting GA4 ecommerce | The GTM template’s Track ecommerce events checkbox translates view_item / add_to_cart / purchase pushes into Personyze interactions. |
| On-page detection | Any site, no code | Point Personyze at the elements: the product id on the page plus the add-to-cart button and order-confirmation signal. Configured in Settings → Recommendations or right inside the wizard’s Interactions step (below). |
| JavaScript API | Single-page apps, custom carts | Log events from your own code at the exact moment they happen. |
| REST API | Server-side and offline events | products_interactions — also the road for historical transactions. |
The Interactions step
Every recommendations wizard carries an Interactions step that shows the same setup as Settings — with live status per event, so you can see views, add-to-carts and purchases arriving as you test:
Category views — feeding interests
Beyond the three product events, tracking category views builds the visitor’s interest profile, which interest-based algorithms and targeting use.
QA — proving events arrive
- Open Live visits, browse your store in another tab, and watch your own view / cart / purchase events appear within seconds.
- The wizard’s Interactions step and the campaign’s readiness panel show per-event arrival status.
- Product IDs must match your catalog — an event whose id isn’t in the catalog can’t attach to a product; that mismatch is the most common “tracking works but recommendations don’t learn” cause.