Collect Data

When you install and configure Product Recommendations, the module deploys behavioral data collection to your storefront. This mechanism collects anonymized behavioral data from your shoppers and powers Product Recommendations. For example, the view event is used to compute the Viewed this, viewed that recommendation type, and the place-order event is used to compute the Bought this, bought that recommendation type.

To learn more about the behavioral data that the Product Recommendations events collect, see the developer documentation.

Healthcare customers

If you are a healthcare customer and have installed the Data Services HIPAA extension, which is included with the Data Connection extension, Product Recommendations stops collecting storefront event data because it is generated on the client side.

To resume collecting and sending storefront event data, re-enable event collection for Product Recommendations. For more information, see General configuration.

Data types and events

There are two types of data used in Product Recommendations:

When you install the magento/product-recommendations module, Adobe AI aggregates the behavioral and catalog data, creating Product Recommendations for each recommendation type. The Product Recommendations service then deploys those recommendations to your storefront in the form of a widget that contains the recommended product items.

Some recommendation types use shoppers’ behavioral data to train machine learning models and generate personalized recommendations. Others rely only on catalog data. To start using Product Recommendations quickly, choose from the following catalog-only recommendation types:

Cold start

When can you start using recommendation types that use behavioral data? It depends. This situation is referred to as the Cold Start problem.

The Cold Start problem is the time required for a machine learning model to train before it can produce effective recommendations. For Product Recommendations, Adobe AI must collect enough data to train its models before you deploy recommendation units. More data generally improves recommendation accuracy and usefulness. Because data collection occurs on your live site, start this process early by installing and configuring the magento/product-recommendations module.

The following table provides some general guidance for the amount of time that it takes to collect enough data for each recommendation type:

Other variables that can impact the time needed to train:

To help you visualize the training progress of each recommendation type, the create recommendation page displays readiness indicators.

While your live site collects data and the machine learning models train, complete the remaining testing and configuration tasks. Once the models have enough data to generate useful recommendations, deploy the recommendation units to your storefront.

If your site doesn’t receive enough traffic (views, purchases, or trends) for most product SKUs, the learning process may not complete, causing readiness indicators in the Admin to appear stuck. Readiness indicators help merchants choose the best recommendation type for their store, but they are only a guide and may never reach 100%. Learn more about readiness indicators. Learn more about readiness indicators.

Backup recommendations backuprecs

When insufficient input data prevents a recommendation unit from returning all requested items, Adobe Commerce fills it with backup recommendations. For example, after you deploy the Recommended for you recommendation type on the homepage, a first-time shopper may not have generated enough behavioral data for personalized recommendations. In this case, Adobe Commerce displays items based on the Most viewed recommendation type.

If input data collection is insufficient, the following recommendation types fallback to Most viewed recommendation type:

Caveats