Speech Analytics Helps Companies Find Meaning in Customer Conversations

A global Internet service provider had a major customer retention problem. The company had spent years looking for a process or system to reduce customer attrition, but no solution worked. The ISP still did not have a good handle on the reasons why its customers were closing accounts. And, therefore, the organization's leaders could not fix the problem.

In 2006, executives decided to conduct a three-month trial of speech analytics to see if they could find actionable insights that would help them curb attrition. The company analyzed a small portion of its call recordings: 25,000 hours of call-center audio—or 150,000 calls.

Managers were surprised by what they learned from the speech analytics application. For the first time, they were able to precisely identify the reasons why customers cancelled their accounts. While some of the reasons were expected, others were totally unanticipated.

Executives at the ISP learned that the company's technology platform and fundamental service issues were a large part of its problem. Once they began to address these issues, their retention rate increased for the first time in many years.

Structuring the unstructured
This is a good example of how speech analytics can dramatically enhance customer service. Speech analytics applications take unstructured conversations, structure them, analyze the content to find meaning and then recommend appropriate actions. Once conversations are structured and analyzed, enterprises can use the data for their own purposes, as well as for the benefit of their customers.

While a few companies around the world have succeeded in creating a data repository that can track and analyze all customer activities, the vast majority of businesses have still not reached this important goal. As a result, customers are often frustrated when companies do not appear to "know" their needs or "care" about them. Likewise, customer-facing staff-in sales, marketing and service are often placed in uncomfortable situations because they lack information—data that customers expect them to have that was shared with another department in the same company.

The global ISP used the results from its speech analytics application to identify agent training needs and to enhance their retention scripts. It also identified a number of easy-to-fix technical issues that were causing a large volume of their cancellations. As a result of the initiatives, the ISP increased its save rate by 1.35 percent, representing more than $1.3 million per year of saved customer revenue.

Speech analytics at a glance

The pilot cost $200,000, a significant expense. But executives knew that if they reduced customer attrition by even as little as 1 percent, the application would have a payback of less than three months. So, while there was risk associated with this pilot, the potential gain made it worthwhile.

The primary use of speech analytics today is to gain insight into the underlying reasons why customers call; this is also known as "root cause analysis." Once company leaders know the underlying reasons for call traffic, they can utilize this information to address the issues. Speech analytics is being used successfully for many purposes, such as increasing the first-call resolution rate, reducing call volume to contact centers, recognizing agent training needs, identifying new product ideas and sales opportunities, reducing customer attrition, discovering operational issues, detecting fraud and ensuring that agents adhere to scripts by giving required disclaimers.

Speech analytics provides a new and valuable way to gain insights into customers' needs and wants. Today it is largely being employed to give enterprises a timely view of customer issues. In the future it will be used to analyze and classify every customer contact, enabling companies to take action when and where appropriate.

DMG Consulting's 2007 Speech Analytics Market Report, a 270-page guide to the speech analytics market, addresses the market, vendors, competitive landscape, technology, product functionality, accuracy, ROI, market share, pricing and implementation best practices. And it includes eight case studies.