What is Text & Sentiment Analytics?

Text analytics is the process of deriving structured data from unstructured text. It helps you find patterns and topics so you know what customers are thinking about your brand, product, or service and allows you to act on trends before they start to become larger issues or impact more customers.

 But how does text analytics actually work in CustomerGauge?

Text analytics involves both topic and sentiment analysis. 

  • Topic analysis categorizes customer comments into business-relevant topics. For example, “our account manager is really supportive” would be categorized under the topic of “Account Management”. 
  • Sentiment Analysis allows you to determine the sentiment - positive, negative, neutral, or mixed - of a customer comment. One of the best uses of sentiment analysis is the ability to get to the “feeling of a comment” without having to read every single comment. 


  • Text analytics is intended for customers with more than 6,000 comments per year.
  • The initial Text Analytics Library requires:
    • A minimum of 5% comments (based on your annual number of comments) per Topic, each with a minimum of 20 words.
    • A minimum of 5 Topics, and a maximum of 15 Topics.

How can you get started with Text & Sentiment Analytics?

Based on your industry, our Text Analytics functionality provides you with an industry-specific training library such as IT, Retail, Financial Services, etc. As soon as there is a completed survey with a comment, the CustomerGauge system will automatically categorize those comments into topics for you. That way you can focus on analyzing your results and improving the customer experience based on the most emerging topics.

For example, available topics in the IT Library may include:

  • Product
  • Account Management
  • Support
  • Implementation
  • Business value
  • Training
  • Price
  • Sales Process

We are constantly monitoring and improving our industry-specific training libraries by adding new data to it and measuring our neural network classification accuracy. Moreover, our models can be further tailored to your particular company by adding your clients' comments to the library with the relevant topic(s) assigned. This will help train your system's library and algorithm and make it more specific to your business. 

Here’s a visual example of how the CustomerGauge system automatically categorizes comments into topics following a survey completion:

1. The survey is submitted with a comment in the comment box

2. The comment box carries the following text: “Have been having problems with the tool lately, particular functions do not work.

3. Based on your industry, our functionality will automatically categorize “product” as topics for this comment.

4. You can then quickly analyze topic results and their frequency in the CustomerGauge system - that way you can easily understand which topics are emerging and need your attention. 

5. To take your analysis to the next level, we’ve also provided the option to link NPS and Topic frequency - the image below shows the topic of “Product”, its frequency in customer comments (36), and the NPS of these 36 surveys (33). In other words, from these 36 survey comments that were assigned the topic "Product," the overall NPS of "Product" from these comments was a 33.  The further to the right the topic dot is, the more it was mentioned by promoters. 

6. You can track your sentiment analysis here as well. Sentiment analysis will automatically register this as a “negative” sentiment and save your team time from having to read the entire comment or ticket to understand if something needs to be acted upon quickly. This way your product and support teams can more effectively and efficiently monitor customers dissatisfaction and stop it in its tracks before it escalates or turns into a public customer review.

View the entire list of reporting options and how to setup here. 

Enrich the Reporting Labels with Your Terminology

We also understand that customers in various businesses may use more diverse terminology than our standard industry libraries offers. For example, instead of using “Account Management”, some businesses may use “Customer Success Team”. As a result, we’ve made it easy for you to provide us with the Reporting Labels you’d like to use for topics without affecting the accuracy of results, just the name given.

Turn on Keyword Recognition

We realize that your customers might know your products and services in more details and they may provide more specific feedback in the comment box.

For example, instead of saying :

“ Your product is great, but I would prefer for your support team to respond faster.”, a customer might say:

“Your Delivery Manager is great, but I would prefer for your CSM team to respond faster.”

To make sure you are fully supported with text analytics, we’ve also added an option to set up Keyword Recognition - you can provide us a list of relevant keywords you would like to see tagged along with the topic (view example below).


Note: Please be careful with keyword matching as it is string-to-string matching. For example, if you choose a word that is frequently a part of other words, any of those words would cause that topic to appear for that comment. For example, if you added the keyword "fin," with a topic of "sharks," you'd run into the following situation where all of these comments would be tagged as "sharks":

"The fins on my shark costume were too small" --> this one is fine, but then you have the following that would all get tagged as "sharks" as well

"My Account Manager has done a fine job of keeping us up to date on the latest product developments."

"I finally received my package in the mail, but only after 2 months!! What's up with that?"

"I'd really like to be able to define my own email settings."

"I really like that you refined the items on your menu. It's much easier to navigate now!"

"Our team is under a bit of financial stress right now, so I'm not sure if we'll continue to have room in the budget for this project."

And so on. So again, please be mindful when setting up keyword matching. 

Keeping your Library updated

Keeping Library updated:  If you need to make improvements or updates to your library, your CSM can help you with this. 

  • For adding new topics or new example comments for existing topics to your library, please send them an Excel with the example comments in the first column and an example topic in the second column.
  • If you need to assign more than one topic to the same comment, just list that comment again on the next row in your Excel with another topic. 

Making Corrections: If you want to make corrections to existing comments to help train the library for the future, your CSM can help with that as well. 

  • Please provide them an Excel with the comments from your platform, the corresponding CGIDs (number_customergauge), and which topics should be added or removed. 
  • If you think any comments were correctly assigned in the first place, your CSM can also help you "promote" those, so that the system becomes even more likely to identify similar comments in the future and apply the same topic.