Adding an attribute where each record has a unique value, such as SKU, to your personalization strategy is not recommended as they are unlikely to generate meaningful personalisation signals.
When configuring your strategy's facets for Classic Personalization or your attributes for affinity for Advanced Personalization, it’s important to choose facets that:
Are strong indicators of user affinity (for example, category or brand)
Are shared across multiple records
Personalization relies on repeated interactions with the same facet values to build affinity scores. If an attribute has high cardinality (meaning most records have unique or rarely shared values), it won’t be effective because:
- Users are unlikely to interact with the same value multiple times
- Algolia cannot build a reliable affinity signal from it
- The attribute may dilute the impact of stronger signals in the model
For best results, select facets that group records into meaningful clusters where repeated user behavior reflects clear preferences.