Suggested reading:
What is a "Legacy Strategy" and should I update it?
A "Legacy Strategy" refers to recommendation settings configured directly inside a campaign before the Reusable Recommendation Strategies were introduced. You should update these to Recommendation Strategies to make them reusable across multiple campaigns and to centralize your management. You can do this by clicking the "Update" option in the campaign design step.
Why does the "New Arrivals" algorithm show old products?
The algorithm uses the date the item was first added to the product catalog, not the current date. If you re-upload or re-integrate an old product that was previously in the feed, it may retain its original creation timestamp.
Why do "Viewed Together" and "Purchased Together" sometimes show unrelated products?
These algorithms learn which products are commonly viewed or purchased together within the same session. If users frequently view or buy two unrelated items in the same session (e.g., a shirt and a coffee mug), the algorithm learns this connection. To ensure thematic consistency, you can add a Page Context Filter (e.g., category + matches the item they're currently viewing) to keep results within the same category as the base product, or anchor the algorithm on a specific behavior using Recommend Based On so the base product itself is more deliberate.
Why do I need to wait to use the Chef Algorithm?
The Chef algorithm uses machine learning to test and predict the best-performing algorithm mix for your specific audience. It requires 60 days of data (learning from the first 30 days and measuring results in the next 30) to accurately train the model.
When I test "User-Based" algorithms in Incognito, why do I see generic results?
User-Based recommendations rely on a unique UserID and browsing history. In Incognito mode, you are a "new user" with no history. The system automatically falls back to a backup strategy (usually Viewed Together or Most Popular) to ensure products are still displayed.
What is the difference between "Most Popular" and "Trending" products?
Most Popular: Ranks products based on total page views over the lookback period.
Trending Products: Calculates a trend score based on the increase in views/purchases over the last 7 days compared to the previous week, highlighting items gaining viral traction.
What is Attribute Affinity?
Attribute Affinity personalizes recommendations by prioritizing products that match a user’s historical preferences. You can select up to 5 attributes from the Product Attribute page to be used in the affinity calculations. If a user frequently visits "Red" items, the system boosts red products in their recommendations.
Can I exclude specific products from recommendations?
Yes. You can configure exclusions in the Recommendation Strategy Settings, or you can exclude specific products completely from all of your recommendation campaigns by setting the Product Activation Status as passive from the Catalog Manager.
Does a filter also apply to the fallback algorithm?
Yes, static and page context filters always apply, including to the fallback algorithm.
For User Context and User Attribute filters, this depends on the “Ignore User-Context Filters” setting: if enabled, the filter is skipped when the shopper has no matching data, rather than being enforced.