Debunking NLP: Analyzing Customer Effort and Emotion

By: Ellen Loeshelle

December 14, 2017

Clarabridge Analytics
Customer Effort
Artificial Intelligence
Machine Learning
Natural Language Processing
Natural Language Understanding

This blog post is part of our Debunking Natural Language Processing (NLP) series. Throughout this series, Ellen will highlight several features that help Clarabridge users go beyond simple topic analysis. This series will show you how new types of analysis aren’t so farfetched after all!


“In Whoville they say that the Grinch’s small heart grew three sizes that day.”  – Dr. Seuss


In late 2015 and early 2016, the digital world got sentimental. A place which had been predicated on instant gratification and click rates suddenly had a change of heart – quite literally. Our favorite digital platforms started embracing their feelings. Twitter replaced the star with a heart to indicate a favorite tweet. Facebook supplemented their iconic thumbs up with a heart and four other faces to indicate a wider array of feelings. Slack followed suit as well by giving users the ability to react to any message with hundreds of different symbols and shapes. These new symbols afforded us the ability to “respond” in a way that much closer matches how we feel.


In the CX world, sentiment is our thumbs up or our star. It provides an arbitrary quantification of the polarity of the experience(s). At first, it was our best available mechanism to quantifying how positive or negative a customer’s experience was, but we continue to hear today that sentiment is not intuitive and many organizations struggle to communicate its meaning to their stakeholders and decision makers. Overtime, sentiment has become a crutch for understanding the customer. It’s ubiquitous in CX applications and demos; red and green word clouds or bar charts haunt us like the Ghost of Christmas Past. However, like the old social media icons, sentiment, while a good barometer for the quality of an experience, may be insufficient to express the subtleties of a customer’s true feelings.


Simon Sinek, in his famous TED talk, points out that the inner two sections of our brains (the Limbic brain) are responsible for generating our feelings, behaviors and decisions. Interestingly, it is this area of the brain that has no capacity for language. So, it’s no wonder that we often struggle to articulate how we feel about our experiences. It’s certainly unfair, then, to attempt to quantify a customer’s experience simply by looking at positive and negative words in their survey response or their chat dialogues. In order to really digest these experiences, we need tools that help to categorize and quantify our emotions and the level of effort that we exert when interacting with brands and services. In the past, sentiment has been the proxy for this kind of analysis. We can take a giant step forward by using emotions and effort as the clues that help us truly understand the impact that we have on our customers’ limbic systems. (Read: how likely they are to be a brand advocate, to purchase, or to churn!)


Our toolbox for analyzing customer experiences continues to grow here at Clarabridge. In 2016, we introduced a CX specific Emotions model. It’s important to realize that not all emotions are created equal in CX. In our world, “grief,” is less common and less actionable than “confusion” or “surprise.” Our Expanded Emotions model offers nearly 50 distinct emotions that help CX professionals align customer feedback to specific and actionable feelings. In 2017, we introduced a new model for Effort that gives analysts the tools to segment which areas of their product or service yields different tiers of easiness or difficulty to customers. Our customers use these tools to simplify complicated and confusing parts of their products, to mitigate anxiety regarding service offerings and to capitalize on excitement and ease in product marketing. We love this new movement toward emotion and effort. Unlike sentiment, which we’ve had to teach to our employees and customers year after year, the reaction that we get to emotion and effort analysis is that it just makes sense. Period. We’ve spent our whole lives mastering interpreting and responding to each other’s feelings. We don’t need to rely on mechanisms that we invented simply for the purpose of putting numbers on a page.


We believe that this is just the beginning of our journey into analyzing and designing CX with an empathetic bent. Our lab is busy prototyping other tools and techniques for analyzing both emotion and effort that will give Clarabridge users more humanism in the way they interact with data. Make no mistake, solely focusing on sentiment is focusing on the effect and missing the cause. Language tells a much richer story when we can analyze it in multiple dimensions and we are so excited to see what stories our customers find by leveraging the new tools for emotion and effort analysis in our toolbox.

Want to read more on emotion and effort? Check out this blog post from 2016:


To read the previous blog posts in this series, please visit:

Debunking NLP: Introduction

Debunking NLP: Translation

Debunking NLP: Named Entities

Debunking NLP: Detecting Products, Brands, Companies and Industries

Debunking NLP: Detecting People, Phone Numbers and Email Addresses

Debunking NLP: Detecting Actionable Sentences