Skip to main content
Gainsight Inc.

How Text Analytics Works

Using Surveys, you can gather information regarding customer pain points, parts of the product that are working, and suggestions that can drive action.

The Text Analytics feature uses several analytical techniques to break down and classify the textual information contained in all text-based questions and for all questions that have comments enabled. It then makes the analyzed information available for reporting purposes. To use this feature with Surveys 1.0, you need to be enable it within your survey settings. For more information on how to complete these steps, see the article How to Enable Text Analytics. This feature is enabled by default for Surveys 2.0. For more information on how to view Text Analytics within Surveys 2.0, refer to Surveys 2.0 Analytics.

Below is an explanation of the techniques used by this feature:

Summarizing Customer Opinions

The most important information contained within a survey comment is the opinion expressed about specific topics. Customers express their opinion about product features, the onboarding process, sales process, support tickets, etc. It is very important to identify the topics and the sentiments expressed on those comments. The opinion expressed can be broadly categorized into three buckets: positive, negative or a suggestion.

Identification of opinion comments and the associated sentiments is referred to as Aspect based context summarization

Aspects are the topics customers have expressed an opinion about in a particular sentence. The survey comments are not analyzed as a whole chunk of text, rather comments are categorized into different sentences. Topics are identified in the sentences and the category of opinion is identified. Comments are split into sentences as one comment can have multiple topics and corresponding opinions.

Examples:

Aspect

Sentence

Sentence Category

customer

Just started using the software at my company over the last couple of months and it has made managing my customers a lot easier.

Positive

customer

Continue to listen to customers about their real-world day-to-day use and needs from your software.

Suggestion

renewal process

The contractual renewal process has been disappointing, time consuming, and frustrating.

Negative

 

Even a sentence holding opinions can have multiple topics:

Aspect

Sentence

Sentence Category

Gainsight product

The Gainsight product is good, but we've had to work around a lot of bugs / quirky behaviors while working on our deployment.

Positive

deployment

The Gainsight product is good, but we've had to work around a lot of bugs / quirky behaviors while working on our deployment.

Negative

Note: Comments with less than five words will not be considered for summarization.

Grouping Aspects

The Text Analytics feature will also group together aspects it identifies as related, and will document this classification within the Aspect Group field of the Salesforce object GSSurveyTextAnalytics.

Note: For users that had enabled this feature when it was known as NPS® Text Analytics, the original GSNPS®TextAnalytics Salesforce object will still be used to capture NPS® Text Analytics data to prevent disruption to existing reports.

Many aspects can be logically grouped together as they refer to different facets of the same topic. For example, the aspects gainsight, gainsight product, experience with gainsight, culture gainsight talk about different experiences with gainsight and can be grouped together under that Aspect Group.

Aspect Group

Aspect

Sentence

Sentence Category

gainsight

gainsight product

The Gainsight product is good, but we've had to work around a lot of bugs / quirky behaviors while working on our deployment.

Positive

gainsight

experience with gainsight

And that is exactly what seemed to have been missing from my experience with Gainsight.

Negative

gainsight

behalf within gainsight

I feel both advocate on our behalf within Gainsight.

Positive

gainsight

culture gainsight

John Doe does not have the same customer-centric culture as Gainsight and that makes for a more challenging mission.

Negative

Displaying Text Data

For Surveys with text questions or comments enabled, there is a section in the Survey module that displays the Text Analytics results.

Snip20170901_3.png

This section of a Survey has a question drop-down field. Select a question and the analytics data for the associated answers will be displayed. The analytics data contains three sections: Opinion, List View, and Word Cloud.

Opinion: This will show a break down of the percentage of Positive, Negative, and Suggestion comments associated with the question.

List View: This will show all aspects sorted by sentence count. The top section will display a summary view showing total number of sentences, as well as a breakdown by sentence category. The lower list view will have a column for Aspect Group, with the individual associated aspects in parentheses, as well as columns for the count of sentences within each Aspect Group, and an Aspect Sentiment column that shows the sentiment of the sentences containing each Aspect Group.

 

Word Cloud: This displays the top 50 Aspect Groups. Hover over each aspect group for additional details.

Snip20170804_12.png

Note: The Text Analytics section will display the message “No Data Found For This Survey” if there is no text analytics data found. This will usually occur if:

  • The survey has 20 or fewer Survey Responses with comments.
  • After processing the survey, no opinion sentences were found.
  • Although Gainsight supports multilingual surveys, the Survey Text Analytics for an individual gainsight survey is supported only for the english text feeds. In other words, non-english free text responses won’t be considered for Survey Text Analytics.
    Note that Text Analytics for multilingual surveys is supported in CX Center.

If you have questions or feedback about the feature explained in this article, please share them on community.gainsight.com.

NPS, Net Promoter, and Net Promoter Score are registered trademarks of Satmetrix Systems, Inc., Bain & Company and Fred Reichheld.
  • Was this article helpful?