Hello. My name is Paul and I’m an analyticoholic. There, I said it. I’m an analytics junkie and I’m not afraid to admit it. Ever since I started writing about performance analytics in 2003, I’ve been hooked. Over the years, my dependency has gotten worse as new strains of analytics found their way into the market.
When speech analytics hit the market a decade ago, it wasn’t that big a deal. I figured I could handle it, especially in its purest form. I didn’t believe all those stories about performance analytics leading to more hard-core analytics solutions. I only dabbled in speech analytics. It never became a habit. In fact, I recall the 2008 survey of end-users that Saddletree Research conducted along with the National Association of Call Centers (NACC) at The University of Southern Mississippi. Over 70% of respondents indicated that they had no interest in speech analytics for their contact centers.
How dangerous could analytics be?
Then someone slipped me a taste of text analytics. I mean, it was just for social media, right? Who cared what was going on with Twitter and Facebook as customer communications channels? I could handle it. But then it happened. Someone talked me into using text analytics to mine for intelligence in the mountains of archived emails in the contact center and I started to think that I may have an analytics addiction. But it wasn’t out of control.
Not too long ago, I wrote about what I only half-seriously referred to as predictive analytics. I wrote that predictive analytics would be able to tell us the effect of what is going to happen before it actually happens. Well, guess what Buck Rogers—predictive analytics is here.
I met John Caddell at a financial services conference that was being held near my office in Scottsdale, Ariz. John is the executive director of Nexidia’s financial services practice. Since I am interested in understanding how speech analytics is being applied in other parts of the enterprise outside of the contact center, like finance and financial services, I arranged to meet John while he was in town.
In the contact center, speech analytics is used primarily to track and understand customer behavior once that behavior has occurred. For example, a contact center analyst might use speech analytics to look for specific words or phrases that are spoken every time another specified word, such as “cancel,” is part of a customer conversation. This context can be exceptionally helpful in a number of enterprise functions beyond customer service, such as marketing, to better understand and effectively address consumer behavior.
Predictive analytics takes speech analytics to the next level. It is proving to be critically important in market segments such as financial services. As John Caddell explained to me, “Predictive analytics is the science of evaluating many types of data to understand why customers do what they do, such as cancelling an
account. Before they cancel, the customer has usually left a trail of activities that can indicate why it happened. Perhaps they had some issues that were never resolved, or they had a contact from a competitive provider. Predictive analytics searches through all the interaction data and finds predictive patterns that can help companies such as banks understand who might cancel before it actually happens.”
Predictive analytics has the potential to alter the course of a transaction while that transaction is in progress, before the outcome is determined. Drawing upon traditional speech analytics processing, archived data is sorted and interrelated factors and/or triggers that influence a particular action are identified. These predictive factors as identified in the analytics evaluation process are then used as required to trigger an action that can potentially change the outcome of a transaction or conversation.
“We’ve built models that we run against a company’s incoming calls,” Caddell continued. “The system alerts the appropriate people in the company to customers at high risk of cancelling. The company can then proactively contact the most valuable of their high-risk customers, work through any issues that may exist and incent them to stay. The result is a reduction in cancellations which usually translates to a multimilliondollar annual benefit to the company.”
Predictive analytics also has the potential to positively impact fraud prevention. For example, fraudsters often use consistently specific techniques in their attempts to commit fraud. If these patterns are identified in real-time during the course of a call, appropriate action can be taken to prevent fraudulent activity.
“Ultimately, we’ll be able to apply these predictive rules in real-time for scenarios like fraud,” Caddell explained. “In these cases, calls that fit the predicted fraud pattern can be subject to some additional validation or inserting a step that a fraudster wouldn’t pass, such as a verification code sent by text message to the customer’s mobile phone.”
For analytics junkies like me, predictive analytics represents another temptation in a long line of increasingly tantalizing offerings. I take comfort, though, in knowing I’m not the only analyticoholic out there. According to the Saddletree Research/NACC 2013 year-end survey of endusers, speech analytics once again tops the list of technology solutions that will be evaluated for purchase or have been funded for purchase in 2014. In fact, four of the top five technology solutions that will be purchased in 2014 are analytics solutions.
There is no relief in sight for analytics junkies. Analytics solutions continue to evolve and improve, becoming increasingly attractive and making it impossible to kick the analytics habit. And to think I used to worry about getting hooked on phonics.
Paul Stockford is Chief Analyst at Saddletree Research, which specializes in contact centers & customer service.
– Reprinted with permission from Contact Center Pipeline, www.contactcenterpipeline.com