Text Analytics Meets Speech AnalyticsText Analytics Meets Speech Analytics

Mining the information in contact center calls using speech analytics can be an early warning system, before an issue escalates to negative social media.

For several years speech analytics solutions have been available. In the contact center space, the technology refers to analyzing and categorizing recorded phone conversations between companies and their customers. Useful information can be discovered relating to strategy, product, process, and operational issues. A mix of large (Verint, NICE) and smaller (CallCopy, Utopy) vendors offer speech analytics.

With the heightened interest in the commercial uses of social media, a perhaps less-well-known technology has increased visibility: text analytics. Text analytics involves lexical analysis to study word frequency distributions, pattern recognition, etc., turning text into data for analysis with natural language processing (NLP) and analytical methods. Again, the goal is to "mine" text for information that can help a company improve customer service, get reactions to new products or policies, etc. Text analytics can help companies turn 100,000s of tweets, Facebook comments and blog posts into actionable data.

Last week speech analytics solution provider Verint announced the addition of text analytics to its portfolio. Partnering with Clarabridge, Verint now offers as part of its Impact 360 Suite a single solution that can aggregate analytic data from both voice recordings and text-based sources.

The graphic below does a great job of showing how the comments on a site like Travelocity can be broken down to reveal, not just an overall rating but an analysis of the various attributes of the comment. While the consumer in this case might have given a typical 2 or 3 star rating, text analysis shows that he/she hated the bed but was very pleased with the valet parking.

The assignment of numerical data to the comments is useful for consolidating data from multiple sources and customers to identify trends, problems, opportunities, etc. An uptick in bed complaints on the heels of beginning the deployment of a new type of mattress, for example, could be flagged this way. Tweets or Facebook statuses could be analyzed the same way; though those text strings might be shorter, they too could involve multiple sentiments that could be rated separately.

One point made by Daniel Ziv, VP for customer interaction analytics at Verint, highlights how the current social media craze could bring broader benefits in customer care. Because of the attention social media can get (think antennas on the iPhone this summer), companies right now are more willing to look at tools like text analytics to help them get a handle on it. Given the chance, what Verint will explain is that mining the information in contact center calls using speech analytics can be an early warning system, before an issue escalates to negative social media. The high visibility of social media may help draw attention to the call recording data that can offer an even richer insight into the voice of the customer.