Sentiment Rules Update

We are introducing Sentiment Classification Rules, a new configuration feature that allows you to customize how sentiment is scored across both Listening and Smart Inbox/Care workflows. The new rules experience, located in Global Settings, provides more control by supporting message-level overrides to Positive or Negative sentiment based on keywords, phrases, or emojis, as well as the ability to ignore specific keywords from being considered by the sentiment model. This functionality is now available to all customers on the Advanced Plan and above, and applies to Listening, Care, and Reporting messages.

Note: This feature is available to customers on the Advanced Plan+

Features

• More control: overrides Sentiment at the message level vs. just influencing scoring
• More expansive within Sprout: applies to Listening, Care, and Reporting messages (if customers are in Reporting Sentiment beta)

Main capabilities:

1. The ability to override messages to positive Sentiment if they contain a word in the Set Sentiment to Positive list.
   1. E.g., messages with “rad" are always considered positive
2. The ability to override messages to negative Sentiment if they contain a word in the Set Sentiment to Negative list.
   1. E.g., messages with "explosion” are always considered negative.
3. The ability to ignore keywords from being considered by the model if they are included in the Ignored Keywords list. 
   1. E.g., "cancer” is always ignored. “I am a cancer patient who loves ACME outpatient center." becomes, to the model, “I am a <> patient who loves ACME outpatient center."