Personalize product recommendations
Choose the right recommendation signal, fallback, merchandising rules, and presentation for email, web push, popup, and workflow content.
What you’ll achieve
- A product module that remains relevant for known customers and complete for low-history recipients without showing empty or stale items.
Postępuj zgodnie z przepływem pracy
-
1
Verify Data → Own Data → Products
Confirm that active products have current titles, images, prices, availability, variants, and destinations. Recommendation quality cannot exceed the synchronized catalog quality.
-
2
Choose a behavioral or commercial source
Use Recently viewed, Similar to recently viewed, Similar to recently added to cart, Similar to recently bought, Purchased item for replenishment, Similar to replenishment item, Bestsellers, or curated Static products according to the campaign intent.
-
3
Match the source to the moment
Browsing signals suit discovery and consideration; cart signals suit recovery; purchase similarity suits cross-sell; replenishment suits repeatable goods; bestsellers and static selections suit broad or low-data audiences.
-
4
Always configure a fallback
Choose Bestsellers, Recently added, or Static products for recipients without enough history or when a selected item is unavailable. Check that the fallback itself has valid products.
-
5
Control quantity and selection
Set the maximum number of products and, for Static mode or fallback, select the exact items. Use fewer, more relevant items when the message has a single objective.
-
6
Customize the presentation
Where the channel supports it, set columns, image, name, description, price, button visibility and label, typography, colors, card background, border, radius, spacing, padding, and hover treatment. Keep touch targets and mobile stacking readable.
-
7
Preview identified and fallback states
Test once with a contact who has suitable browsing or purchase history and once with a low-history recipient. Verify product availability, price, currency, image, and destination at render time.
Verify your work
- A behavior-qualified test contact receives the intended recommendation source
- A low-history test contact receives the configured fallback
- Every rendered product has a correct image, title, price, availability, and destination
If something doesn’t look right
No products render
Confirm the source is not Do not show products, synchronized products are active, and a fallback exists.
Recommendations look generic
Use a test contact with attributable product events and check pixel identity.
Wrong price or stock
Wait for the connector sync and verify the source product record before sending.
Limitations and safeguards
Some channels expose fewer visual controls than email, and behavior-based personalization depends on identifiable, consent-compatible history.