REVOLUTIONIZING CONTACT CENTER QUALITY MANAGEMENT WITH SPEECH ANALYTICS TABLE OF CONTENTS Introduction...1 Challenges of the Traditional Quality Management Process...1 Challenge #1: Small Sample Size...1 Challenge#2: Creating Objective and Reliable Evaluation Criteria...2 The Solution: Analytics-Driven Quality Management...3 Objectively Defining Agent Skills using Speech Analytics...3 Reliably Evaluating Every Conversation with Speech Analytics...3 Conclusion...5 Introduction Most contact centers these days record some or all of their inbound and outbound telephone calls in order to be able to assess the quality of the customer service being delivered, as well as to ensure that agents are following any required procedures. In most contact centers, the most efficient way to achieve these objectives is to optimize agent performance, thus improving the customer experience and legal compliance, all of which directly affect an organization s bottom line. Therefore, the primary focus of most contact center quality measurement processes is to measure (and hopefully to improve) agent performance. Challenges of the Traditional Quality Management Process Traditional Quality Management (QM) programs generally rely on a person, normally a supervisor or QM Analyst, listening to recorded calls and manually scoring the events in each call on an evaluation form. Due to its manual nature, this is a very time-consuming and expensive process. Challenge #1: Small Sample Size Because it is such an expensive process, contact centers generally can only afford to evaluate a small fraction of their calls, often a sample rate of less than one percent of the calls occurring within the contact center, which is hardly statistically reliable. Furthermore, this small sample is often randomly selected. As a result, managers are often unsure if the results of the monitoring session accurately reflect the agent s performance. This may cause them to hesitate to act on the information they receive. It also allows agents, in some cases with good justification, to challenge the results as not representative of their normal performance. Another consequence of the low sample rate is that changes in performance are difficult to spot. So, if an agent receives coaching or training on a specific area, it is hard to tell from a single call monitoring session if the agent responded with behavior changes. While it can be argued that the solution to the low sample rate is to evaluate more calls, this is normally unworkable because of cost. It is rare to get budget approved to increase the sample rate, and if approved, such funds are very vulnerable to cuts.
Revolutionizing Contact Center Quality Management with Speech Analytics / page 2 of 5 Challenge#2: Creating Objective and Reliable Evaluation Criteria Another challenge with traditional call monitoring is creating performance categories that can be reliably measured. The definitions of performance categories may be vague, making measurement difficult. Any subjectivity in the definitions of performance opens the door to evaluators using their own judgment on how to assess a call, creating disagreement between evaluators and requiring costly calibration sessions to insure that all evaluators score subjective criteria the same way. Performance categories may also be multidimensional, so if the agent performs well on one sub-topic in the category but poorly in another, what score does the evaluator give? To illustrate this challenge, below is an example of how a major U.S. company defines several areas of its call evaluation process: Customer Engagement > Maintains a friendly and helpful tone throughout the call > Stays focused and attentive to the customer > Demonstrates active listening and responding Builds Loyalty > Conveys goal to be the customer s preferred vendor > Motivates customer to use product > Acknowledges customer loyalty > Attempts to resolve customer concern in one call > Attempts to prevent future problems > Takes ownership of call > Recaps call to confirm solution Under the Customer Engagement category, how would the evaluator determine if the agent had a friendly and helpful tone? What specific behavior does the evaluator look for? Or, is it a judgment call, opening the door to disagreement between evaluators? Furthermore, what happens if the agent had a friendly tone through only half of the call? Is the agent given credit for the skill or not? If the evaluation form uses a five-point scale, does the agent get a zero, a two, or a four? Ambiguity in the measurement criteria will generate chronic variability in the performance scores that is completely unrelated to the agent s actual performance. Looking at the same quality dimensions listed above, under the category Building Loyalty, what happens if the agent acknowledges customer loyalty but neglects to motivate them to use the product? Do they get a passing score on the category or not? Also, how would an evaluator determine that an agent has attempted to prevent future problems? We see from this example the difficulty in creating an evaluation form that insures that all evaluators, whether supervisors, QM Analysts, or the agents themselves, hear and evaluate the call in exactly the same way.
Revolutionizing Contact Center Quality Management with Speech Analytics / page 3 of 5 The Solution: Analytics-Driven Quality Management Speech Analytics can resolve these challenges with traditional Quality Management (QM) by automatically measuring what agents do on calls, reducing and perhaps even eliminating the need to listen to and evaluate calls manually. Using Speech Analytics to drive Quality Management offers several important advantages. First, Speech Analytics automatically measures all calls, so the concern about sample size is eliminated. This means that everyone, whether supervisor, manager, agent, trainer, or HR Representative, can trust that the results represent the agent s typical behavior. Second, for Speech Analytics to work well, the events it measures must be behaviorally defined. In other words, events are defined by the specific things that the agent says, not by general concepts of agent behavior that are open to interpretation. This forces companies to become much more disciplined in how they evaluate agents. Furthermore, if a QM category is multidimensional, such as Building Loyalty above, the system can measure and report on each discrete skill, so the agent gets credit for what they do well, and specific deficiencies are also revealed. Objectively Defining Agent Skills using Speech Analytics For example, in a recent study conducted for a client interested in automating the Quality Management (QM) process, one category on their evaluation form was Presenting Offers. Part of the definition of that category was that effective sales effort was shown by the agent. The challenge here is defining the meaning of effective and effort. To resolve this, Genesys defined the skill behaviorally, as: Did the agent offer Product X? All of the phrases that the agent could say when offering the product, such as: We have X Product, or Have you considered Product X?, were defined for the Speech Analytics system, so that the system would automatically identify when the agent used those phrases and objectively measure each agent s usage of that skill. If the agent offered the product, they were given credit for performing in this category. If they did not offer the product, they did not get credit. Reliably Evaluating Every Conversation with Speech Analytics A critical pre-requisite to this type of analysis is the ability to reliably recognize entire phrases said by agents during conversations with customers or prospects. Unlike every other Speech Analytics product, which begins by converting the speech into text or phonemes, then searches within the converted speech, the patented Speech-to-Phrase Recognition delivered by Genesys Speech Analytics uses an iterative phrase recognition algorithm applied directly against the audio itself. This algorithm produces results that are much more accurate and complete than any competing Speech Analytics algorithm, especially when analyzing topics such as agent skills, which must be defined by groups of phrases. Additionally, Genesys Speech Analytics and Analytics-Driven Quality Management automatically monitor every agent s usage of the key skills during every conversation on an ongoing basis, alerting supervisors and the agents themselves when performance needs improvement. This level of specificity and reliability is crucial for accurately measuring agent performance and can paint a very different picture from what the company gets using traditional methods. For the company above with the QM category Building Trust and Loyalty, their traditional QM assessment process gave the agents a high average score of 84%. However, when each of the components in that category was defined as phrases in Genesys Speech Analytics and measured discretely, average agent scores for Building Trust and Loyalty dropped to 29%.
Revolutionizing Contact Center Quality Management with Speech Analytics / page 4 of 5 Build Trust and Loyalty AVERAGE AGENT SCORE USING ANALYTICS-DRIVEN QUALITY MANAGEMENT AVERAGE AGENT SCORE USING TRADITIONAL QUALITY MANAGEMENT Customer Appreciation 45% Preferred Card 22% Encourage Usage 20% Average 29% 84% Add Value Mention Cash Back 43% Mention Sweepstakes 24% Mention Automatic Payment Plan 13% Mention Website 23% Mention Company Newsletter 13% Mention Customer Account 13% Average 21% 43% Table 1: Comparing Average QM Scores from Traditional QM versus Analytics-Driven QM In another category, Add Value, that involved offering products and value-added services, traditional QM scored agents at 43% while Genesys Analytics-Driven QM gave them less than half that rating with a score of 21%. Again, the term Add Value is relatively vague and open to interpretation, while Genesys Speech Analytics, powered by its patented Speech-to-Phrase Recognition, was able to objectively measure specific discrete skills such as: Mention Cash Back, Mention Sweepstakes, etc., as illustrated in the table above. These skills were then defined by the specific phrases the agent would use when leveraging that skill, which produces a uniform, objective measurement system for every agent on every call. In this case, Analytics-Driven QM was able to show that the results delivered by the traditional QM process were exactly the opposite of what the agents were actually doing. Thus, Analytics-Driven QM quickly identified a major opportunity to significantly increase customer retention and revenue.
Revolutionizing Contact Center Quality Management with Speech Analytics / page 5 of 5 Conclusion Speech Analytics can revolutionize traditional Quality Management by measuring all calls, not just a small sample, and helping companies define Quality Management criteria in a way that significantly improves the objectivity of measurement. This dramatically increases the reliability of assessment and the trust everyone has in the results, enabling companies to act on findings rather than debate their validity. This precision allows the kind of research not possible with traditional Quality Management methods, often illuminating significant revenue opportunities and identifying the impact of specific skills and practices that optimize performance. Companies can track performance gains, or the lack thereof, immediately after a coaching or training session. They can use that information to assess the willingness and ability of the agent to improve performance, the quality of the training effort and the success of the company in supporting that effort, and the transfer of skill from the training environment to the operations floor. Because the process is automated, a side benefit is that the cost of quality management can be significantly reduced, and QM personnel can focus on specific areas of opportunity or work with agents to improve performance rather than just randomly measuring how they are doing. The result can be a dramatic improvement in agent performance, often with significant financial impact, achieved very quickly. Corporate Headquarters Genesys 2001 Junipero Serra Blvd. Daly City, CA 94014 USA Worldwide Inquiries: Tel: +1 650 466 1100 Fax: +1 650 466 1260 www.genesys.com About Genesys Genesys, the world s #1 Customer Experience Platform, empowers companies to create exceptional omnichannel experiences, journeys and relationships. For over 25 years, we have put the customer at the center of all we do, and we passionately believe that great customer engagement drives great business outcomes. Genesys is trusted by over 4,700 customers in 120 countries, to orchestrate over 24 billion contact center interactions per year in the cloud and on premises. For more information visit: www.genesys.com, or call +1 888 GENESYS. Genesys and the Genesys logo are registered trademarks of Genesys Telecommunications Laboratories, Inc. All other company names and logos may be trademarks or registered trademarks of their respective holders. 2013 Genesys Telecommunications Laboratories, Inc. All rights reserved.