Edition: Enterprise

TeamSupport is leveraging IBM’s Watson technology to provide internal sentiment analysis on Ticket Actions. You may notice public Actions scored with sentiments like satisfied, polite, and frustrated, along with scoring on the Ticket.

The goal for the Watson data is to help you better understand your relationships with your Customers so that you can continue to provide them with exceptional service.
Additionally, you can use the service to understand how your team’s written communications may be perceived by your Customers. This analysis can highlight training opportunities for your agents in order to improve the tone of your communications and elevate conversations with your Customers.

Ticket Sentiment is available as a weight in our CDI calculation. The two values working together will enhance your understanding of your customer health and satisfaction.

Going forward, we are planning to add Ticket Automation rules which will let you run triggers on Action sentiments.


The following sentiments are used:

  • Positive emotions: Satisfied, Excited, Polite
  • Negative emotions: Sad, Frustrated, Impolite, Sympathetic

The tone detector is ‘tuned’ specifically for support interaction, and will detect language that fits the listed emotions from the context of a Support Agent/Customer conversation. Tone is expressed by your word choices, punctuation, and the level of formality in your writing.

Here are some example sentences for each sentiment.


  • Yes, that time works fine for me.
  • Excellent, that worked.
  • I appreciate the help with this.


  • That’s awesome, thank you!
  • I’m glad we were able to fit you in so quickly!
  • I’m glad that solution works for you.
  • I am happy to help.


  • Please let me know your next availability.
  • Thank you for your patience.
  • May I suggest using our reference guide?
  • I’d like check on a few things first.
  • Could you please provide me with your ID#?


  • I am very disappointed with those options.
  • I really wish that hadn’t happened.
  • We feel like our concerns are not being heard.


  • This has been a very frustrating experience.
  • This is not acceptable.
  • I don’t like the options presented.


  • This is ridiculous.
  • Your whole company is useless.


  • I’m sorry you’ve been having trouble logging in.
  • I apologize for the inconvenience.



The text of each public Action on a visible Ticket is analyzed for sentiments. Actions created by Users and Contacts are analyzed. The result may be that no sentiment was detected, or there may be one or more sentiments detected. If detected, the sentiment(s) will be listed at the top of the Action along with a percentage score which illustrates the confidence that the Action has the indicated sentiment. For example, text that is analyzed with the results of “Frustrated 66% Sad 74%” indicates with a 66% confidence that the author of the text was expressing frustration, and 74% confidence that the text expresses sadness.

You may not see this information instantaneously as it may take a few minutes to populate.

Scores 50% and above are only listed to ensure that the system is providing a confident score. When evaluating scores on Actions, it’s important to realize that 50 is the lowest calculable score. With this perspective, a score of 57%, for example, can be considered a “low” score.


Public Actions on visible Tickets created by Contacts are used to provide the Ticket with a sentiment score. The score is calculated by averaging the highest confidence sentiments of each of the Contact’s Actions. Using the above example, the “Frustrated 66% Sad 74%” Action would be interpreted as a Sad 74% Action because 74% is the higher of the two scores on that Action. Since Sad is considered to be a negative emotion, it is averaged in as a negative value (-74). If there is an additional Action with a positive sentiment such as Polite 55%, that would be averaged in as a positive number (+55). Thus the average of the two would be (-74 + 55)/2 = -9.5. The Ticket scores are expressed with a value in between 0 and 1000 where:

  • Negative: 0-499
  • Neutral: 500
  • Positive: 501-1000

Therefore the number is scaled as follows: (5 * -9.5) + 500 = 453 (rounded up). The result of 453 indicates that they system is more confident that the Ticket has negative sentiment (< 500) than positive.

Agent Training

Ticket Sentiment Analysis provides many opportunities for growth within your Support team. Here are a few focus points for agent training:

  • Did the Customer start out frustrated at the beginning of the Ticket, but their tone improved by the end of the Ticket? How did the agent’s tone affect the tone of the Customer?
  • Did the Customer’s tone turn from positive to negative? Could the Ticket have been escalated before it turned sour?
  • Are there any patterns in the tones of successful agents?


Ticket Sentiment Analysis can be disabled at any time in the Admin panel.

Need more help with this?
Customer Support

Was this helpful?

Yes No
You indicated this topic was not helpful to you ...
Could you please leave a comment telling us why? Thank you!
Thanks for your feedback.