Last week I was looking at a demonstration of one of the analytics tools that we work with. The demonstrator was breaking down call dispositions and cross-referencing them with things like the type of call, the eventual final destination of the call, and even things like which language the person chose (English or Spanish). Using the tool, he showed me an interesting piece of information: the percentage of people that hung up on the automated voice system was insanely high for people coming into the Spanish system (almost 50%) as compared to people coming into the English system. It turns out that our consultants had found this, and investigated further. The result? Aha! The "native Spanish speaker" that was speaking the prompts for the Spanish system had a Cuban accent. Most of the people calling in to this particular system were of Mexican descent, I think because the call center was operating in the southern United States. I was further informed that "some Mexicans don't get along as well with Cubans" and that this cultural clash was causing people potentially to hang up earlier than expected.
My thought? Holy crap, there's a lot of subtleties involved in keeping a contact center's automated system operating smoothly. This was a great example of how various analytics tools can really dig down deeper than the normal statistics and log reports that are often generated and associated with most typical automated voice systems.
It also emphasized a greater point. The tool itself could not have picked out that particular problem. Someone had to know where to look. And even then, the tool only alerted the people that were working on the optimization of the systems that there might have been an issue. It was then bringing their expertise to bear that identified the problem and enabled us to solve it.
It seems like most of the contact centers that we at Nuance are dealing with today are inquiring about analytics. Often, however, they're asking about the tools themselves -- will it analyze my recorded conversations? What sort of reports will it generate? How quickly can it analyze X calls? While these can be valuable questions to ask for any particular tool, they're not the questions that a contact center should be asking when considering adding analytics to their analysis of the contact center's performance. First they should be thinking of the problem they're trying to solve: are they worried about their automation rate? Customer satisfaction? Whether agents are complying with scripts? Once they’ve identified what the actual problem is, then they can determine which tool is best to solve it. Individual tool vendors often think of their tool as a hammer and every problem as a nail. But the only way to truly give the customer the best possible experience is to look at the analysis of the contact center's performance with multiple tools that are designed to solve a particular problem. That's our approach and so far it's met with success.