Loading

How Data Checker considers the data in Einstein Recommendation Builder.

Date de publication: Jun 3, 2026
Description

Within the Einstein Recommendation Builder (ERB), the Data Checker tool displays two columns: Record Count and Required. These columns show whether your org has sufficient data to train an Einstein recommendation model.
The minimum required record counts for the objects used in the Einstein Recommendation Builder are:

Minimum Record Count Requirements

ObjectMinimum Required Records
Recommended Items10
Recipients100
Positive Interactions400

If your record counts fall below these thresholds, the Einstein Recommendation Builder (ERB) will not be able to generate a recommendation model. This article explains how these record counts are calculated.

Résolution

How Record Counts Are Calculated

When setting up the Einstein Recommendation Builder (ERB), you select and associate three types of objects:

  • Recommended Items — the objects you want Einstein to recommend (for example, products or articles)
  • Recipients — the objects that will receive recommendations (for example, contacts or users)
  • Interactions — the objects that record how Recipients have engaged with Recommended Items (for example, purchases or clicks)

The Interactions object must have either a Lookup or Master-Detail relationship with both the Recipients object and the Recommended Items object.

Positive Interactions Count

The Positive Interactions record count reflects the total number of records in your selected Interactions object. Einstein Recommendation Builder requires at least 400 Interaction records for the model to train successfully.
For example, if you selected the Interaction custom object, the Data Checker counts all records in that object. If the count is below 400, the Data Checker will show insufficient data.

Recommended Items Count

The Recommended Items record count reflects the number of Recommended Items objects that have at least one related Interaction record. In other words, only items that have actually been interacted with are counted — items with no interaction history are excluded.
Einstein Recommendation Builder requires at least 10 Recommended Items that have related Interaction records.
For example, if you selected Recommended_Items as your Recommended Items object, and Recommended_Items__c is the lookup field on the Interaction object pointing to Recommended_Items, the Data Checker counts only those Recommended Items records that appear in at least one Interaction record.
Note: The Recommended_Items__c field referenced above is a custom Lookup or Master-Detail relationship field on the Interaction object, relating it to the Recommended Items object. Your field name may differ based on your org configuration.

Recipients Count

The Recipients record count reflects the number of Recipient objects that have at least one related Interaction record. Similar to Recommended Items, only Recipients with interaction history are counted.
Einstein Recommendation Builder requires at least 100 Recipients that have related Interaction records.
For example, if Recipients is your Recipients object and Recipients__c is the lookup field on the Interaction object pointing to Recipients, the Data Checker counts only those Recipient records that appear in at least one Interaction record.
Note: The Recipients__c field referenced above is a custom Lookup or Master-Detail relationship field on the Interaction object, relating it to the Recipients object. Your field name may differ based on your org configuration.

Numéro d’article de la base de connaissances

004052218

 
Chargement
Salesforce Help | Article