You are here:
Item Recommenders
An item recommender presents items from your catalog based on each user's engagement history. An item can be a product, an article, a property listing, a course, a service, or any other entity you serve to users. The item recommender is the default recommender configuration for personalized recommendations across web, email, Agentforce, and other channels.
How Item Recommenders Work
An item recommender uses DLPR with the engagement signals you collect for your items and your item catalog. As users discover, evaluate, and convert items, DLPR builds a journey-level understanding of each user, and ranks items in a way that reflects both relevance to the user and alignment with your business goal.
Engagement Signals
An item recommender learns from all interactions associated with an item. The specific signals depend on your industry and the type of item you're recommending. In general, signals fall into three categories:
- Discovery interactions: How users find and view items, such as item views, impressions, and category browsing.
- Consideration interactions: How users evaluate items, such as clicks, add-to-cart, save-for-later, time-on-page, and shares.
- Conversion interactions: How users commit to items, such as purchases, redemptions, subscriptions, bookings, applications, or inquiries.
- Custom signals: Any signal tied to your business objective that isn't covered above.
The signals you map to the recommender are the most important input decision you make. They define what DLPR treats as a positive engagement.
Model Inputs
In addition to engagement signals, an item recommender uses these inputs.
- Item catalog metadata: Titles, descriptions, categories, and any attributes that describe your items. For a product catalog, this includes brand, price, and SKU. For an article catalog, this includes author, topic, and publication date. For a property catalog, this includes location, bedrooms, and amenities.
- Objective configuration: The business goal that a recommendation optimizes for.
Guidance for Your Catalog
A rich, well-structured catalog improves recommendation quality, especially for new items. Following these guidelines can help improve recommendation quality in personalized output.
- Provide complete titles, descriptions, categories, and attributes for every item.
- Keep catalog metadata current as you add new items.
- Configure engagement signals that reflect your objective and the nature of your items.
