Mike Ashe has a philosophy about CRM: “If we’re right more often than we’re wrong then we’re going to beat random.”

By that, he’s referring to the analytics behind his company Mattersight's Predictive Behavioral Routing which “predicts which (customer service) agent is best-suited to handle the next call based on individual customer behavioral data and agent performance data.”

Ashe, vice president of Chicago-based behavioral analytics software provider Mattersight, was one of the speakers for the recent CXPA webinar -- “Personality Matters” -- highlighting Mattersight’s approach to customer retention through behavioral-based interactions. The concept, he states, is simple and straightforward:

  • Classify each consumer’s behavioral characteristics
  • Analyze how each employee performs with different customer types
  • Use the behavioral pairing analysis to assign work/content

Behavioral pairing involves something that resembles match-making -- using the content and language of previous calls from clients to determine their basic personality types, then analyzing which types each customer care representative (CCR) communicates best with in order to reduce costs and increase service levels. Beyond that, in sales environments the approach, according to Ashe, can “see a high impact on customer retention rates of 30 to 50 percent.”

Four Personality Types

Those basic personality types have come from 10 years of tracking and tagging customer behavior through a data repository of over 70 million customer profiles tied to phone numbers (while retaining the anonymity of the people attached to them). From this lengthy project they’ve broken personalities down -- similar to, but simpler than, a Myers-Briggs testing -- to four basic types that they claim should make up 95 percent of calls to contact centers:

  • Emotions-based
  • Thoughts-based
  • Opinions-based
  • Reactions-based

As an example of the pairing process, Ashe presented data from a pilot program at CVS Health Customer Care -- one of their recent clients and success stories -- with statistics for Eugene, a CCR with CVS, showing that the best match for his customer service style would be opinion-based callers.


Learning Opportunities

Predictive Behavior Routing

The pilot program ran for 60 days at CVS’s San Antonio site using 80 agents -- including Eugene -- compiling comparable information. Two-thirds of calls were routed with Mattersight’s Predictive Behavioral Routing, with one-third serving as the control group, randomly routed on their own.

Like a good matchmaker, the calls that found their way using that combination of customer behavior tagging and data from tracking calls through the customer care site were paired with the agents best suited to respond to individual personality styles.

The results after only 30 days into the program, according to Gloria Walker, the director of customer care at CVS Health, were “compelling enough for us to make the decision to move it out across all of our customer care sites to full production.”


ATT = Average Talk Time       MATR = Mattersight

While the general transience of cell phone usage wasn’t addressed in the presentation, Ashe seems confident in the foundations of the predictive behavioral routing system. “We will never be absolutely certain of anything. But we’re using data that can make the best approximation and as we typically keep reminding ourselves, all we have to do is beat random. “