Real Time Speech Analytics in the Call Center
Illustration by David Gladis

Do you listen to your customers in real-time? What types of things do you need to know about what’s happening now versus reviewing post-call analysis? A major challenge for many contact centers is determining which calls to live-monitor to identify potential issues before they escalate—especially when you need to meet critical requirements, be they compliance, business procedures and/or customer satisfaction.

Of all the useful contact center technologies available today, one of the most valuable I’ve seen in addressing these challenges is real-time speech analytics. While the call is in progress, supervisors or analysts can receive alerts based on the detection (or lack of detection) of key phrases required of agents or phrases that indicate a customer issue. The alert criteria may be based on the detection of a specific phrase uttered by the agent or customer, or an overall scoring threshold based on an accumulated scoring of several detected phrases.

Of course, just as WFM requires dedicated, real-time WFM analysts to be effective, speech analytics requires dedicated, real-time speech analysts to understand when intervention may be warranted (e.g., whisper coaching or chat to agent). This can be one of the most powerful benefits for supervisors who have spent countless hours live-monitoring calls in hopes of identifying the ones where there is potential risk and where agents may need assistance.

Real-time guidance, a feature often combined with real-time speech analytics, is a great tool for improving the performance of your frontline staff. It empowers agents by putting at their fingertips the information that they need to respond to customers, reducing the need to seek assistance or transfer calls.

Tip: Don’t distract your agents by bombarding them with a constant stream of popups, tips and alerts. If you’re considering this feature, focus on what is truly critical, has potential risk, or can increase agent responsiveness to your customers.

Is every agent a high performer? Not likely. Our centers include agents of varying tenure with different levels of communication skills and knowledge. Listening to and comparing the calls of your high performers to those of lower performers can help to identify specific skills or training gaps. You can use this information to guide your decisions about what to detect with real-time speech analytics, and how best to use real-time guidance to increase performance levels.

Real-time also provides post-call value

Real-time solutions capture key phrases as the call is being recorded, so there is no delay in processing time to search recordings. In addition to real-time insights, results add post-call value to help focus improvement efforts. For instance:

  1. The ability to quickly identify risk, process or compliance issues. Regulatory compliance is a well-defined problem. Analyzing call recordings with speech analytics is an effective way to identify and quickly address compliance issues. Focus training and coaching efforts by searching within a speech analytics application for keywords and phrases that would reveal non-compliant calls, offending agents and use of prohibited terminology. Results can provide clear examples for agent training and coaching sessions.
  2. The ability for a supervisor to identify “good” calls. Yes! There are tons of good calls that need our attention, too! Speech analytics can also be used to identify positive words or phrases during a call; for example, when a caller expresses his or her thanks to an agent for being especially helpful. These calls serve as powerful training aids that can help to improve agents’ problem-solving skills. They are also important positive reinforcements that can bolster morale and provide greater incentive for good performance.
  3. The ability to target the right calls for monitoring. Contact centers have long struggled to determine which calls to monitor and evaluate. Do you pick the 10 longest calls? Do you always select calls from Monday, because that’s your peak call volume day? Or maybe afternoon calls because that’s the highest right-party-contact (RPC) timeframe? Real-time speech analytics can offer far more relevant criteria for call monitoring. For instance, how about monitoring every call that included the phrase, “I’m going to sue you” (or any variation of it)? And if that phrase was uttered by an agent, it can be sent to his or her supervisor. If it was spoken by the customer, it can be flagged for follow-up by the legal team. You can see how speech analytics would dramatically fine-tune call monitoring so that supervisors can address the right calls with minimum effort—with the end result being more effective and efficient evaluation efforts while avoiding hefty fines.

Have your quality improvement, evaluation and compliance monitoring efforts been helped by other speech analytics capabilities? If so, I’d love to hear about them! Please send me your comments.


Lesley Vereen is a Senior Product Manager at Genesys, where she provides product vision and development direction for PureCloud WFO quality management solutions. Lesley’s extensive domestic and international contact center experience includes contact center management, consulting and training across various industries. She has led initiatives to improve customer experience, quality monitoring and coaching programs, contact center management and frontline customer service training.