The explosion of digital and social media has changed the way we communicate – as people, as consumers, and as companies. While this has made hearing the voice of the customer easier than ever, it has also introduced a new challenge: too much information (or, as the Twitter Generation would say, TMI).
Much of this content is unstructured text in comments or social media postings. The challenge is to chisel out the “gold” nuggets trapped inside this bedrock of customer feedback and use the refined raw material to make decisions aimed at specific business goals.
In my chapter “The Power of Babble: Using Text Analytics and Data Mining for Uncovering Actionable Customer Insights” in Allegiance’s book Delivering Customer Intelligence, I explain how the right combination of text analytics and data mining can be a company’s best friend in weeding through all this information and finding out what the customer really thinks and wants.
Text analytics using natural language processing (NLP) can analyze text as spoken or written by human beings. NLP is sophisticated enough to discern facts and entities (such as persons, places and things) as well as attitudes and opinions.
For example, in analyzing a call center note, such as “The system is slow,” NLP analysis would identify a thing (system) and an associated qualifier (slow). Creating a pattern such as “System +” would ensure that similar statements, though worded differently, would be identified and categorized.
Typically, the data collected from text analytics is used alongside structured customer survey and transactional data such as customer satisfaction scores, geographic data, demographics, purchase and usage histories, product-feature data, etc. But the challenge is determining which unique combinations of facts extracted from verbatim comments, structured survey scores and demographics have a measurable impact on customer loyalty or satisfaction.
In traditional structured data analysis, data mining has been an effective tool to find those unique patterns that identify sub-groups of customers who are more likely to be loyal or disloyal. Through this data mining, we can identify and refine patterns and trends among hundreds, even thousands of variables. We can then make predictions based on information obtained by analyzing and exploring this data.
Using advanced text analytics, we are in essence creating more discrete variables that can be used to create more meaningful data mining models. For example, in the area of customer loyalty, data mining, when preceded with text analytics, can determine which facts and concepts extracted by the text analytics engine have the biggest impact on loyalty scores or satisfaction ratings. These new data mining variables become extremely powerful because many times emerging issues will be first expressed in a verbatim comment.
For example, imagine a pattern that showed an extremely loyal group of customers in a specific geography who scored high on product quality and mentioned a new product feature in their verbatim comments. This would give product managers and marketers early information about where to focus messaging and additional product development.
Data mining and text analytics are both advanced analytic methodologies that significantly improve an analyst’s ability to find hidden insights to drive measurable change in customer experiences. Combining them creates an extremely powerful method to discover actionable insights in a fraction of the time it takes using more traditional methods. These results can then be quickly acted upon to achieve specific business goals for improving operational efficiency, customer engagement and product innovation.
Eric Weight is Vice President, Solutions Consulting, at Allegiance