Voice Search Technology Fills the Future of Contact Centers with Challenges

SAN DIEGO—With the onset of more viable mobile Web services, experts at the Voice Search 2009 conference predict that call center traffic will increase.

Sitting on the panel “How Will Contact Centers Evolve in the Voice Search Era?,” Monique Bozeman, principal analyst at Monique Bozeman Consulting, says the ease with which users can move directly from the mobile Web to a given channel within a call center will drive the numbers.

“Where are voice search queries going to land?” asks Joe Benzel, chief marketing officer at Speech Cycle. “They don’t stop at the cell phone.”

One of the immediately obvious consequences of heavier call volumes is the strain placed on call center resources. Benzel says he’s seeing some firms trying to deal with this by moving to a pay-for-agent model, where callers pay for an interaction with a live agent and shoulder part of the economic burden. He has seen some traction for that model in Europe, but points out that it has some significant drawbacks.

“Pay-for-agent puts a burden on call centers to make sure their tech is good,” he says.

“That trend that they’re paying in Europe is disturbing. I don’t know how many interactions with live agents I’ve ever wanted to pay for,” quips Bozeman, drawing a handful of chuckles from the room.

“In a tough economy, there are call centers that are just trying to answer calls,” she adds.

Bozeman contends that even as technology advances, its implementations are not consistent. She points to a “continuum of sophistication” in call centers. Many are overloaded and comprised of agents lacking the talent or training to handle calls in a smooth way. She maintains that as long as contact centers continue to pay employees around $25,000 per year, they won’t see any real improvement. Annual salaries of $50,000 to $60,000 are needed to draw the right talent.

Bozeman also believes that the infrastructural problems of call centers are artificially keeping call numbers and call duration low. People aren’t calling about all their needs because they are frustrated. She argues that the recognition technology is strong, probably strong enough to handle most call center needs. She places the blame for call center shortcomings on implementations—especially the logical flow of dialogs.

“Forget if recognition is 100 percent,” she says. “When we’ve done analysis on some of the largest firms, you’d be amazed what the business analysts have come up with. Ninety percent of the time the logic was wrong. It had nothing to do with the technology.”

She goes on to cite an example: At an unnamed call center where the points of departure—those points where frustrated callers hung up before completing transactions—were monitored, it was found that 99 percent of the departures took place at the same point in the dialog. When users were asked to give their phone number, they tried repeatedly and failed. Recognition aside, the phone number was unnecessary and played no part in completing the transaction. It wasn’t even needed for verification. Most calls were failing for an entirely unnecessary reason. If design were taken more seriously and streamlined, the argument follows, resources and money could be saved.

Pointing to another way of cutting costs without cutting service, Bentzel asserts that a software-as-service model may be ideal in a number of situations. He admits, however, that there has been significant resistance to adoption.

“The reality is that a shift for customers in payment plans makes them think they must be getting a bad deal,” Bentzel says. “Software-as-a-service model is breaking through that, but plenty of companies that have said, yes, they would be okay with success-based pricing, go back to the traditional because, over time, they think it will cost less.”

A steady drumbeat of calls, however, may force them to reconsider.