Limitations

Limitations of the product recommendation system.

General limitations

Product recommendations are powered by Google AI and have the following inherent limitations:

All items in a product catalog must use the same language and currency.
Google AI projects are limited to using a single product catalog for training AI models. For customers with multiple domains using different sets of products/categories, this will result in cost and management implications as additional Google AI projects, catalogs and AI models will need to be trained and maintained. Due to this complexity, support for multi-domain use is currently not supported.
Product IDs tracked in your Fanplayr integration must be consistent with the IDs used in your product catalog.

Only one Recommendation component can be used in a widget.
A widget that displays product recommendations can only use a single language. If you set up multiple catalogs with different languages in your account you will need to create a distinct widget for each language that uses product recommendations.

A minimum number of user events must be captured in order to produce high-quality AI models. Please check the AI Models section for more information regarding data requirements and limitations for each model type.
The initial training of each AI model is estimated to take 7-14 days depending on the size of the product catalog and the amount of traffic generated by your store.

Only product displays that use Ranking models can use the product list feature. Displays that use an AI recommendation models must always use the full product catalog.

Last updated 2 months ago