Item recommendation analytics

Learn about item recommendation analytics and how to view them in Braze.

Viewing analytics

You can view analytics for your recommendation to see which items users were recommended and how accurate the recommendation model was.

1. Go to Analytics > Item Recommendation.
2. Select your recommendation from the list.

Available metrics

Audience

These are metrics related to your recommendation audience, which includes precision, coverage, and recommendation type.

Refer to the following table for more information:

Metric	Description
Precision	The percentage of time the model correctly guessed the next item a user purchased. Precision is heavily dependent on your specific catalog size and mix, and should be used as a guide to understand how often the model is correct. In past testing, we have seen models perform well with precision numbers ranging from 6-20%. This metric updates when the model next retrains.
Coverage	What percentage of available items in the catalog are recommended to at least one user. You can expect to see higher item coverage with personalized item recommendations over most popular ones.
Recommendation type	The percentage of users who will receive personalized or most recent recommendations versus the fallback of most popular items. The fallback is sent to users who don’t have enough data to generate a personalized or most recent recommendation.

Items

This table includes metrics about your personalized, most recent, and most popular items from your catalog.

Refer to the following table for more information:

Metric	Description
Personalized items Most recent items	This column lists each item in the catalog in descending order of most often recommended to users. This column also shows how many users were assigned each item by the model. Either Personalized or Most recent items will be listed depending on the recommendation type.
Most Popular items	This column lists each item in the catalog in descending order of popularity. Popularity here refers to items in the catalog that users interact with most often in the entire workspace. Most popular is used as the fallback when personalized or most recent cannot be calculated for an individual user.

Overview

This is an overview of your chosen recommendation configuration, which includes when the recommendation was last updated.