Speech analytics has done a great deal to popularize the notion that the contact center can be an important source of enterprise knowledge, according to Jim Davies, a Gartner research director. "The downside is that it is difficult to build a business case for it," he says. "You don't necessarily know what you're going to find until you find it."
Broadly, speech analytics represents a category of technologies with the ability to transform call recordings into clear, concise, and searchable reporting on words, concepts, and even emotions expressed during a customer call-although each product is slightly different. The business case is often initially made through manual call recording reviews, which can provide insight on unexpected customer preferences or reveal previously underrated key competitors. Automate that process, the thinking goes, and the company will be even better off.
In every call, customers not only complete a transaction, but they offer cues and clues to what drives their behavior, and that is exactly the kind of information sales and marketing are charged with collecting and understanding. "There's a tremendous amount of information you can get from the quality system, either from agent note fields, or from the big push to use voice analytics," says Joe Galvin, president of contact center consultancy J Galvin & Associates. For example, "what's the take rate, and what's driving the take rate?"
Not that every deployment of speech analytics is strictly revenue-oriented. It can replace manual, error-prone processes used to monitor customer trends, such as agents keeping a manual tally upon hearing certain key words or phrases. "The only way before speech analytics that marketing got that information was to orchestrate very deliberate feedback through focus groups, or rely on those agents ticking off items on a piece of paper," says Anna Convery, senior vice president of marketing for Nexidia.
NICE Systems says it has an insurance customer using the emotion-detecting capabilities of speech analytics, pairing hostile tones with word detection of legal threats. "They run those reports twice a day, flag the calls, and listen to them to determine what action to take," says Robin Schaffer, senior product marketing manager.
These emerging technologies have their challenges, including the very real problem of making sure that the words being analyzed are reported in their correct context. "If you are a customer with a mobile phone company and you are recorded talking about dropped calls, you could be saying something positive or negative," says Colin Shearer, senior vice president of market strategy at SPSS. The solution to this problem depends on the analytics in place-some attempt to filter by the emotion or tone of voice detected in the voice, while others try to pull in additional contextual information to determine if the company is being praised or knocked for the consistency of its service.
Finally, keep in mind that like any analytical tool, speech analytics are merely a means to an end, and that real change will have to come from multiple departments working together to improve the customer experience. The technology, CallMiner CEO Jeff Gallino says, "is not going to solve your churn problem or your upsell problem alone."