Recommendations

Set up
Create a catalog
Build a catalog for any use case—like retail products, articles, or media—and choose up to 10 customer interaction events to train your machine learning model.

Connect
Map the product ID
Link each event to its unique item ID, such as product_ID or article_ID, to give your recommendation engine the context it needs.

Define
Configure schema fields
Align event properties with the suggested model columns. Go beyond product category by adding parameters to increase granularity and refine recommendations.

Build
Design your product-based audience
Start building your audience after syncing your catalog. Think product-first: choose the upgraded item or brand you want to promote and let the model find the best fit.

Calibrate
Select audience size
Set your audience size using pre-populated options or custom sliders. Targeting the top 20% or less helps you focus on users most likely to convert.

Deliver
Choose your destination
Push your audience to downstream channels and customer journeys. With hundreds of integrations, you can activate recommendations anywhere.

Create
Select users
In the Computed Trait builder, choose which users should have the recommendation trait added to their profiles.

Specify
Define recommendation items
Select the item type and the number of items to recommend. Configure this step to generate product, category, or brand-level recommendations tailored to your use case.

Run
Calculate and enrich profile
See how many users match your criteria, then create the trait. Each trait value is saved to the customer’s individual profile and updates automatically as new data comes in, keeping personalization fresh and dynamic.

Exclude
Implement filters
Optionally exclude specific items or event properties from being added to a profile as a recommendation trait, such as products under $50 or items not in stock.