Watson APIs

Watch your tone! IBM expands on Watson’s tone sensitivity for customer service applications

Share this post:

Key Points:
– Tone Analyzer is expanding its emotional understanding to help businesses solve customer engagement challenges
– The new feature, Tone Analyzer for Customer Engagement, detects communication tones in customer conversations, indicating frustration, satisfaction, excitement, politeness, impoliteness, sadness and sympathy. It can also uncover tone from common emojis, emoticons, and slang.
– This insight can help customer service agents and chatbots respond appropriately to frustrated, sad, or satisfied customers.

Learn about Tone Analyzer


It’s crucial for companies to interact with their customers effectively: It can be the difference between making a sale or losing business. But as every manager quickly discovers, communication with customers is complex. Human conversation is rife with subtlety, including tone of voice. In the era of social media, text messages, chatbots and virtual agents – and even emoji-laden messages – it’s getting increasingly difficult for businesses to keep up. Nonetheless, communication is key to business growth. And customer service agents need to craft the best response to address any customer issue.

Today we’re making generally available a new feature within the IBM Watson® Tone Analyzer service – Tone Analyzer for Customer Engagement. The feature allows users to detect seven communication tones in chatbot and customer service conversations, indicating the following emotions:

  1. Frustration
  2. Impoliteness
  3. Sadness
  4. Sympathy
  5. Politeness
  6. Satisfaction
  7. Excitement

It can even identify tone from common emojis, emoticons, and slang.

What makes “Tone Analyzer for Customer Engagement” special?

Tone Analyzer for Customer Engagement is built to understand conversations between customers and brands. Trained specifically on customer support conversations on Twitter, this model can help monitor agent and customer communications, detect anomalies, and highlight opportunities to improve. It can track how tones progress throughout conversations, and note where agents need to be more sympathetic or polite to customers’ situations, or if they should show more excitement when they resolve the issue.

The new feature makes a chatbot tone-aware, enabling it to provide unique responses to frustrated, sad, or satisfied customers. For example, it can respond to sadness with, “I’m sorry you are upset about this problem,” but to satisfaction with, “I’m glad you are satisfied with our service.”

Tone Analyzer for Customer Engagement also understands emojis, emoticons, and slang as indicators of tone, just like the general-purpose Tone Analyzer model.

Learn more about how this model was built.


Who’s using Tone Analyzer for Customer Engagement already?

Since launching this Tone Analyzer version specializing in customer service last April, several early adopters have applied this technology to improving their business’ customer engagement across professional learning, development training, and ecommerce.

Yappy, a sales intelligence platform, uses Watson Natural Language Understanding to help sellers improve conversion rates through improved communication with customers. By integrating Tone Analyzer for Customer Engagement within “Yappy for Business,” Yappy can take this a step further, helping sellers determine their customer’s tone during conversations before replying to SMS, voice, and Facebook messages, email, and online chat messages.

“My customers, including custom embroidery specialists Queenborough.com, use Watson to determine the tone in replies they are about to send to customers via various messaging platforms,” said Marius Dornean, founder of Yappy. “The integration of Tone Analyzer for Customer Engagement enables companies using the Yappy platform to gauge the overall tone and quality of agent interactions with customers for managers to find training opportunities and evaluate the quality of agents.”

Yappy integrates Tone Analyzer for Customer Engagement to understand tones in agent interactions with customers, as indicated by the emojis. Courtesy of Yappy. (Photo courtesy of Yappy.)

KNOLSKAPE, an experiential talent transformation company in professional learning and development, is using Tone Analyzer for Customer Engagement to train customer service employees to improve engagement with customers. KNOLSKAPE tests employees with simulated customer service scenarios and evaluates their responses using Tone Analyzer for Customer Engagement. The tone analysis gives the employees real-time feedback on which areas they’re doing well in and where they need to improve. KNOLSKAPE is launching custom versions of the product with clients from Financial Services, Manufacturing, Retail, and IT.

“The quality of communication between the company and the users is a top priority for businesses, especially an international one like KNOLSKAPE where the majority of communication happens with customer service representatives in English,” said Chaithanya Yambari, director at KNOLSKAPE Labs. “Tone Analyzer for Customer Engagement gives valuable metrics which help in assessing the quality of this communication to train customer service agents to be consistently polite.”

How do I get started?

  1. Test Tone Analyzer for Customer Engagement in the demo OR in our Customer Engagement Analytics Demo, utilizing both Tone Analyzer and the Watson Discovery Service to find insights in contact center data.
  2. Follow our “Getting Started” guide for Tone Analyzer.
  3. Finally, let us know what you think!

Tip: Check out our “Lite Plan” to test out the service at no cost.


Try Watson Tone Analyzer for Customer Engagement today.


Add Comment
No Comments

Leave a Reply

Your email address will not be published.Required fields are marked *

More Watson APIs Stories
August 14, 2017

How Watson AI is helping companies stay ahead of hackers and cybersecurity attacks

Cybercrime is projected to cause $6 trillion in damages by 2021. There is no way to manually read and analyze the huge volumes of unstructured data that security analysts need to process. Watson's AI allows systems to continuously learn by constantly analyzing billions of data points to detect patterns and even predict attacks before they occur, helping clients analyze threats up to 50% faster.

Continue reading

August 4, 2017

AI is the future of cybersecurity – How Watson helps detect threats faster and better protect your organization

Not having complete context and a holistic, real-time view of cyber threats puts an organization at a huge risk. Global ransomware attacks will exceed $5 billion in damages this year, up from $325 million in 2015. Learn how AI is being used to transform the cybersecurity industry and help analysts identity threats and resolve them faster than ever before.

Continue reading

July 26, 2017

Meet the 13-year-old prodigy taking IBM and artificial intelligence by storm

ABC recently profiled 13-year-old Canadian tech prodigy Tanmay Bakshi who started using computers at five, launched his first app at nine, and has been working with IBM's cognitive APIs for a few years now. In 2013 he built "tTables," an app to help kids learn multiplication, a huge achievement for a child who loves to code but is largely self-taught.

Continue reading