Published in General

More Analytics From Your Text Analytics

Structured survey questions are great tools for gathering customer feedback, but what if the real reason customers are unhappy isn’t a question on your survey? Open-ended feedback is potentially the most valuable feedback tool because it gives your customers the chance to talk about the issues that concern them most, even if there wasn’t a specific question on the survey to address it. If you are fortunate enough to collect large volumes of open-ended feedback, the next question is what do you do with all this data?

In the not so distant past, analyzing large volumes of customer feedback meant hiring a team of people who would read through the comments and manually assign codes to each comment based on themes discovered in the process. Manual coding, however, isn’t scalable. With large volumes of data, this process quickly becomes cost prohibitive. Thanks to text analytics, the days of requiring manual coding are gone.

Hurwitz & Associates defines text analytics as “the process of analyzing unstructured text, extracting relevant information, and transforming it into structured information that can be leveraged in various ways.” (The Hurwitz & Associates Text Analytics Victory Index).

A text analytics Natural Language processing (NLP) engine can interpret the subject of the statement and assign a positive or negative sentiment. You’ll know what people are talking about and whether or not they are happy about it.

Advances in text analytics technology continue to occur at a rapid pace, especially in the arena of sentiment analysis—but even the best solutions aren’t turnkey. The real value will come when your organization buys into the text analytics initiative and gets involved with the project.

To get the best results from text analytics, all solutions require tuning. Tuning is where you refine your data and remove any noise irrelevant to your business.

Text analytics tools, like Allegiance’s text analytics partner Clarabridge, offer machine-learning tools to help tune the data. Does Java refer to a coffee bean or a programming language? Does ARM refer to a financial instrument or an appendage? The answer depends on what business you are in. With machine learning capabilities, you can train the tool to tell the difference for your business.

Text analytics guru, Seth Grimes, recommends working backwards. Start with the question you are trying to answer. “What are your business goals? What data and analyses will produce the insights that guide you in meeting goals?” ( The best person to do this type of work is a business owner.

Text Analytics tools are now easier than ever to use, which means you don’t have to be a technical guru to dive into the data. Follow these recommended steps to get more from your text analytics:

  1. Dedicate some business resources as your text analytics champions.
  2. Set aside time for these text analytics champions to be trained on the process tuning categories and sentiment.
  3. Tune the data. Make sure you are asking the right questions, and understand what you will do with the results.
  4. Repeat Step 3 on a quarterly basis. Your Text Analytics initiative should be an ongoing effort. The topics and sentiment voiced by your customers will likely change over time.

By incorporating text analytics into your customer feedback strategy, you will be on your way to discovering more insight quickly and easily.