The 2015 State of the Speech Technology Industry: Speech Analytics
acknowledgment of a problem, perhaps get an apology, and be assured that a solution is in the works. Such insight and follow-up beats being caught off-guard and ill-prepared to deal with ensuing headaches.
Real-Time or Post-Call Analytics
While some may view speech analytics as an emerging technology, many early adopters have gone a step further and are using not just post-call but real-time speech analytics, which, as the name indicates, analyzes speech while a call is in progress. By assessing specific words in a conversation, a contact center agent can be guided to the next-best action, whether it's to consult a knowledge base, escalate a call to a supervisor, offer a discount or coupon, or transfer the call to the retention center.
Fluss calls real-time speech analytics one of the biggest changes in the market in the last year.
"There are two worlds of speech analytics now," she says. "There's post-call, which is what most people are doing, and there's the emerging real-time analytics, which is enabled by much quicker processing power."
According to Fluss, real-time speech analytics began to emerge in 2011, but it wasn't until 2013 that it became workable. "The potential of real-time speech analytics is substantial, as it can alter the outcome of calls and impact the overall customer experience," Fluss wrote in a DMG newsletter. "DMG expects to see substantial investment in real-time speech analytics solutions in the next few years."
Using Speech Analytics as a Watchdog
Providing safe measures to guard against contact center fraud is of paramount concern to customers. "One out of every 2,900 calls to financial institution call centers is fraudulent," said Greg Adams, vice president of product management at Pindrop Security, in a statement.
There are several solutions to stem the problem, including voice biometrics, but some companies may see those solutions as being even more costly than speech analytics. By using key indicator phrases, patterns and trends can emerge—for example, a contact center might see that someone called in eight times to open a new account but the caller offers erroneous information when asked for a piece of information that he should know. In this case, speech analytics can raise a red flag.
"Customers can give very valuable data and you can leverage that," says Matt Lautz, president and CIO at CorvisaCloud. "For example, a key phrase might be, 'What is your Social Security number?' If a person says, 'Hold on a moment,' it might be a problem. What person doesn't know their Social Security number?"
What to Expect
As speech analytics becomes more widely deployed and if costs come down, it is poised for greater traction. However, there's a caveat: Companies may want to keep their expectations in check, and it's up to vendors to present a realistic picture of what speech analytics can and cannot do.
"You need to educate your customers about speech analytics," Kowal says. "Customers may not want limitations placed on the number of words that can be searched. We have to explain how they can get the best accuracy and be able to take action on the data that is collected."
Fluss says she expects greater adoption and is encouraged that over the past 12 months companies have not just been talking about speech analytics data but taking more actionable steps. "Speech analytics is a change agent, but you need to be positioned to apply the findings—organizations are starting to become better at applying the findings and therefore realize the benefits," she says. "Now, we've finally gotten an appreciation of how to use post-call analytics."
In our first State of the Speech Technology Industry issue, we reveal the latest trends and developments in eight market categories.
Companies and Suppliers Mentioned