Making the Business Case for Speech Analytics
I Can't Answer that Question
"I don't know" is not the answer managers want to give when the board of directors asks what the payback will be for a new project. But with Big Data projects, that response is honest and often accurate. Big Data, the practice of mining large sets of data with the hope of discovering valuable business insights and transforming them into potential actionable items, leads to ROI metrics that are seldom clear at the beginning of a project; the potential benefits only become clearer as the project unfolds.
Essentially, Big Data application development is often an iterative process, one that requires patience. "A corporation may start down 19 different tracks before hitting pay dirt on the 20th attempt," says Nick Heudecker, research director at Gartner. As evidence of the uncertainty and the need for patience, a study by Wikibon, an open-source research firm, found that the ROI for Big Data projects is about 55 cents for every dollar invested. In other words, corporations initially invest a lot more than they are getting back.
Given that level of uncertainty, businesses have been looking for speech analytics' low-hanging fruit, areas of the organization that already rely heavily on speech, so theoretically paybacks would be clear and relatively easy to discern. And there's one area that involves a large daily volume of voice interactions: the contact center. One of Nexidia's customers analyzes up to 150,000 hours of audio each day, according to Teri Navin, product marketing manager at Nexidia. Not surprisingly, these business units are where speech recognition analytics has started to take root.
Bringing Order to Chaos
In call centers, speech analytics has the potential to bring order and structure to customer interactions and help businesses shed light on often seemingly random patterns buried in mountains of words. For instance, call centers are constantly trying to speed up call handling, which reduces operating expenses. A common frustration occurs whenever a call is passed from one agent to another. Often, clients have to restate who they are, why they are calling, and the steps they have already taken to resolve the issue.
Speech analytic systems can be designed to pass customer identification from point to point along the resolution path. Metadata is one Big Data component: It outlines data (what the information is and how it is used) about data. By including metadata in customer interactions, call centers move redundant information from system to system, which speeds up resolution and lowers call-handling times.
Getting to the Root of the Problem
Speech analytics can help agents perform root-cause analysis. "Sometimes, customer dissatisfaction has nothing to do with the contact center or even a product but instead a faulty business process," notes Jeff Gallino, chairman and CTO at CallMiner. Corporations are in a constant state of flux, and ripple effects arise. For instance, a wireless carrier enables customers on its Web site to examine how many minutes they've used during the month. The marketing department redesigns the site, and calls flood the contact center. An analysis of the recorded calls reveals that the Web site change produced the rise in calls.
Speech analytics helps agents deal with irate callers. Trigger words, such as angry, cancel, and profanity, move a call from the typical process into the high-risk area. Here, the system activates call muting and screen masking features so that fallout from when an exchange becomes heated does not impact other interactions.
And analytics enables managers to become more efficient. Many companies count on managers to manually evaluate recorded calls, leaving them little time to perform in-depth analysis. Manually sifting through the information is often inefficient, error prone, and scattershot. Informal processes tend to be ineffective and impractical, especially if the information is not used in a timely fashion. Analytics makes the process more scientific and more effective.
Putting Trends in Context
Managers can work to reduce call volume, improve quality, and reduce customer attrition. Using data culled from customer interactions, they identify trends and address operational and strategic issues, like which agents would benefit the most from additional training. Executives can see which customer issues take up most of the agents' time and work on processes to streamline problem handling. Leaders are then able to take action and coach their team members to reduce hold times.
Analytics can help reduce customer churn by helping the organization change its gaze from internal (reducing costs) to external (helping customers with their problems) processes. Typically, agents are judged by average call-handling times, the lower the better. But that measurement may not be accurate. Sometimes, the person who spends a little more time talking with the customer actually does a better job addressing an issue, which ultimately increases customer satisfaction and improves the company's odds of retaining the customer. Analytics help corporations maintain the proper balance.
Companies and Suppliers Mentioned