Considerations When Adopting Speech Analytics


Bookmark and Share

In today’s hypercompetitive, customer-centric marketplace, every enterprise strives to gain competitive advantage through customer loyalty, high customer satisfaction, and low customer turnover. Through a combination of tools, such as analytic technologies, and data-mining techniques and access to real-time data, companies can now place a greater emphasis on customer engagement and satisfaction. Today’s increased enforcement of standards and stricter legal compliance rules have led call centers to take proactive steps to ensure that they are in compliance with regulations through speech analytics. In the realm of analytic technologies, speech analytics is quickly becoming one of the most demanded technologies in customer engagement optimization and the fastest growing technology in call centers.

The history of speech begins with the first speech recognition systems in the 1950s and 1960s, only understanding digits, as shown in Figure 1. During the 1970s, speech interface projects took off under the U.S. Department of Defense's Advanced Research Project Agency (DARPA), which funded speech understanding research programs. Speech recognition turned toward prediction in the 1980s, when speech recognition began to be used in commercial applications in business and the medical field. In the 1990s, automatic speech recognition began to be introduced to the masses, with the first consumer product, Dragon Dictate, and the first voice portal, VAL (introduced by BellSouth). Since the 2000s, mobile voice interfaces have become more pervasive, and, more recently, we have seen innovations for speech interfaces with TVs and automobiles and real-time speech analytic capabilities for call centers. While speech analytics made its appearance in call centers around 2003, the market for speech analytics had grown from 24 customers in 2003 to more than 3.5 million in 2015, according to DMG Consulting data.

Figure 1. History: Speech Recognition Project

Source: Modified from PCWorld; www.icsi.berkeley.edu  “From AUDREY to Siri”; http://repository.cmu.edu/compsci “Harpy, production systems and human cognition”; www.pcworld.com/article/243060/speech_recognition_through_the_decades_how_we_ended_up_with_siri.html

As mentioned earlier, every enterprise strives to gain a competitive advantage through customer loyalty, high customer satisfaction, and low customer turnover. Today’s increased enforcement of standards and stricter legal compliance rules have led call centers to take proactive steps to ensure that they comply with regulations through speech analytics. There are three key objectives when implementing speech analytics: 

  • Prioritizing dedicated resources.Dedicated resources might include the foundation and processes that accommodate the speech analytics system, which then allows it to collect the data from which the company can derive insights. This step includes continual trial and error, identifying cost-saving and time-saving solutions, training, and maintenance, and automation capabilities to gather transactional data.
  • Improving business processes. Speech analytics produce a vast amount of data that could offshoot into an infinite number of call center objectives. It will take management to carefully consider where to focus efforts, in addition to the ROI, whether it's to optimize operations, supplement existing quality assurance processes, or increase training. By processing and analyzing big data, organizations can pinpoint and identify areas for process enhancement and product improvement, as well as opportunities for increased sales and mitigated risk.
  • Hiring the right people to oversee and maintain your investment. By hiring the right people or promoting from within, supervisors can refocus their time from data collection to coaching, training, and planning. Internal company experience will be crucial to building queries or categories that produce relevant data and reinforce the business drivers unique to the organization.

The adoption of speech analytics is expected to increase as companies continue to recognize its value and see it as a tool that can enhance current quality monitoring methods. Through speech analytics, companies can extrapolate not only the content of the conversation, the customer’s needs and wants, but also the sentiment and emotional aspect behind the voice of the customer.

Recommendations and Guidance for Executives

The speech analytics ecosystem consists of single-solution providers. With the implementation of any speech analytics platform comes a caveat: executives and management tend to view speech analytics as a quick fix. However, without a strong blueprint and processes in place, practitioners will not be able to realize returns, thus, causing a further hindrance to the process. This is especially true of speech analytics, which can lead companies to see it as a one-fits-all approach and become overly ambitious.

Speech analytics should be considered for compliance, performance management, and real time capabilities. Customer satisfaction, business process improvement, sales effectiveness, and risk management guidance benefit from implementing speech analytics. The following guidance for value creation, as recommended by Alex Richwagen, corporate vice president of analytics at New York Life Insurance Company (which has built a very strong internal speech analytics program) is recommended for executives who have been using or looking to implement speech analytics.

  • Start by using speech analytics to identify categories of repeat calls, keywords that could define root causes, and emerging trends. The first step is to identify and understand patterns of behavior, to create guidance rules for real-time speech analytics. This then leads to exploring success cases through repeat calls, while at the same time, identifying root causes and emerging trends through keywords related to products, pricing, customer suggestions, and other key attributes. All of this can lead to better call center agent performance, increased customer engagement, customer loyalty, and customer satisfaction.
  • Create a consistent and repeatable process for call center agents to follow using keywords and trends that have been derived from the first step.
    Based on the analysis of the recorded calls, speech analytics will allow for a consistent and repeatable process to be created for call center agents. This will lead to lower customer turnover, higher sales growth, and potential cross-selling and up-selling opportunities.
  • Create insights from data found from winning cases and emerging trends for management to incorporate data in real time. To accomplish these goals, leadership needs to overcome dominant managerial logic and even prepare for new learning curves. While the traditional business method has been to analyze data retrospectively rather than to address issues in real time, in this case, leadership could benefit from prioritizing real-time speech analytics going forward and incorporating data from a fresh perspective.

Insights for speech analytics can be highly valuable and informative. Numerous organizations in various industries are taking steps to implement speech analytics in true real time. Speech analytics is starting to play more of a crucial role in enterprise. The future we face is automation and simplification of speech analytics into many verticals. While speech analytics can be beneficial on a stand-alone basis, its value increases as it is integrated with other complementary high-value processes and applications. 


J.P. Shim, Ph.D., is a member of the computer information systems faculty and executive director of the Korean-American Business Center at Georgia State University.  He is a professor emeritus at Mississippi State University. He received his doctorate degree from the University of Nebraska-Lincoln and completed Harvard Business School's Executive Education Program. He taught at the University of Wisconsin, New York University, and the Chinese University of Hong Kong. He has published a number of books and articles.