Einstein Conversation Insights Sentiment

Hi Claudia, great question! Here is how Einstein Conversation Insights (ECI) determines Sentiment and what configuration options exist: 

How Sentiment is Determined: 

ECI uses a Salesforce-managed machine learning model to classify conversation segments as Positive, Negative, or Neutral. The model uses contextual NLP -- not simple keyword matching -- to assess tone and emotional signals from the transcript. Sentiment is calculated at both the keyword mention level and the overall conversation level. 

Configuration Options: 

1 

. You cannot retrain or adjust the underlying Salesforce sentiment model -- it is platform-managed and updated by Salesforce automatically. 

2. You can influence what is surfaced by setting up custom Keywords and Competitors (Setup, Einstein, Conversation Insights, Keywords). These keywords appear alongside sentiment labels in the dashboard. 

3 

. 

 You can filter the ECI Conversations dashboard by Sentiment (Positive, Negative, Neutral) to surface calls needing coaching attention. 

4. The classification thresholds are not configurable by admins. 

Best Practices: 

- 

U 

se the Sentiment filter in the ECI Conversations dashboard to identify and prioritize negative-sentiment calls for coaching. 

- Combine Sentiment data with Talk Ratio and Longest Monologue metrics for a fuller coaching picture. 

- Reference Salesforce Help for full ECI documentation:

Hope this helps clarify how sentiment works in ECI! 

Mani G 

P 

r 

incipal/Founder