Einstein Engagement Frequency Model Card

Einstein Engagement Frequency Model Card Overview

The Einstein Engagement Frequency Model Card in Marketing Cloud Engagement is designed to help marketers determine the optimal frequency for sending emails to individual contacts, thereby maximizing engagement and minimizing unsubscribes. Here’s an overview of the key aspects of this model:

Model Details

Purpose: The model analyzes each contact’s engagement record to find the optimal frequency for email messages.
Development: Built using matrix factorization and recommendation system algorithms, the model is fine-tuned for each customer using only their data, ensuring personalized insights.
Training Data: The model is trained on a customer’s historical data, ensuring unique optimization for each client without cross-customer data sharing.
Metrics: Performance metrics such as spread, sparsity, and correlation of output scores are monitored to maintain model quality.

Intended Use

Primary Uses: The model primarily supports the Frequency Split activity in Journey Builder, helping to send the optimal number of emails per subscriber, thus enhancing engagement across campaigns.
Out-of-Scope Uses: It is not intended for prioritizing specific emails based on current saturation levels or manually capping send frequencies per campaign.

Relevant Factors and Inputs

Data Inputs: The model uses up to 90 days of historical engagement data, including sends, bounces, opens, clicks, unsubscribes, spam complaints, and timestamps.
Exclusions: It excludes third-party data, demographic data, and specific content from emails.

Output

Frequency Bands: Generates personalized frequency bands for active subscribers, aiming to optimize engagement.
Environment: Deployed within the Marketing Cloud Engagement environment.