What makes the sentiments positive or negative? 



Date Votes
  • Official comment

    Hello Emily!

    Call sentiments rely on a trained classification model that searches the transcript for words and phrases commonly associated with positive or negative sentiments. It then aggregates and ranks all of the identified words and phrases into a sentiment range, which is listed below:

    • 0 to .1 is Neutral (we expect this to be at least 50% of calls).
    • .1 to .3 is Slightly Positive/Negative.
    • .3 to .5 is Fairly Positive/Negative.
    • .5 to 1 is Mostly Positive/Negative.

    Let me know if you have any further questions, and welcome to the community!

  • Ariane,

    What are the specific words and phrases for each sentiment classification?



  • Hi Nick!

    Thanks for your interest in Call sentiments. Unfortunately it’s not as simple as sharing a set list, the AI is constantly learning and re-training itself to produce more accurate results, so the inputs change with time.

    Thanks and welcome to the Community!

  • Ariane,

    Are you able to share some sample phrases for the various sentiment types that would potentially lead a cal to be classified as a certain sentiment?



  • Hey Nick,

    Unfortunately, at this time we aren't able to do so due to the variability across businesses. The best option is to analyze both positive and negative calls in your account and that should help you understand how it's performing for you. Additionally, I do want to say that we've begun work to improve the call sentiment functionality which will highlight specific words and phrases that are being identified as positive or negative. You can expect to see that improvement in the next few months!



Please sign in to leave a comment.

Didn't find what you were looking for?

New post