Speech analytics can be performed on recorded conversations or in real time. The post-call solutions that analyze recorded conversations are more mature than the real-time applications, which only started to enter the market in 2011 and became viable in 2013. However, the potential of real-time speech analytics is substantial, as it can alter the outcome of calls and impact the overall customer experience. DMG expects to see substantial investment in real-time speech analytics solutions in the next few years. If a solution can do real-time analysis, it can also be trained to undertake the post-call process. The reverse is not true, though; if a solution is designed to analyze calls only from recordings, it cannot perform real-time intermediation as well.
Post-Call Speech Analytics
Most of the speech analytics solutions on the market can analyze recorded conversations. The recordings can come from any vendor that can capture and share them; this includes speech analytics vendors, stand-alone recording vendors, workforce optimization providers, unified communications vendors, cloud-based contact center infrastructure vendors, etc.
The primary output from post-call speech analytics is metadata (files of data) in two general forms: phonetic representations of a conversation or a transcript of a conversation. (Without using other recognition capabilities, the first-pass accuracy rate from a speech-to-text solution may be as low as 40 percent.) Speech analytics solutions today go far beyond the initial transcription analysis performed by their underlying Large Vocabulary Speech Recognition (LVCSR) or phonetic engine. All of the speech analytics vendors have made significant efforts to build an application layer that vastly improves the accuracy of results.
Post-call speech analytics delivers tangible value by conducting discovery on recorded audio to determine why customers call. (Post-call analysis typically occurs on a next-day basis.) Post-call speech analytics is highly effective at identifying new and breaking trends. The value of this information increases when the application correlates disparate but related issues. Speech analytics can identify important and useful findings about a company’s products, services, procedures, policies, customers, competitors, and more. It has also proven to be effective at identifying agent performance-related opportunities that, when addressed, can improve productivity.
Real-Time Speech Analytics
Real-time speech analytics solutions analyze calls as they are happening, and deliver some form of actionable recommendations to companies. A growing number of vendors, including CallMiner, Castel, Interactive Intelligence, NICE, and Verint, now offer real-time speech analytics functionality. These solutions analyze conversations almost as soon as they occur. (A slight delay of a few seconds is necessary for the application to analyze a large-enough snippet of the conversation to identify a trigger for action.) New uses for real-time speech analytics are emerging, but some fundamental applications are already known. Real-time speech analytics is being used to identify situations in which agents are not in compliance with their script, guidelines, or standard operating procedures. These solutions are also being used to identify when customers are very unhappy.
When real-time speech analytics is combined with real-time guidance, agents can be told in real time, while the caller is still on the line, how to rectify a situation. For example, if an agent is supposed to verify a caller by asking specific questions but skips this step, a real-time speech analytics solution can send an alert to remind the agent to do the verification.
Speech Analytics Benefits
The benefits of real-time speech analytics have captured the imagination of contact center managers. Some plan to use it to improve compliance with government regulations, others want it to identify bad agents, and still others hope it can be used to reduce customer attrition. Regardless of its uses, and there are many, the power of speech analytics comes from its ability to alter the outcome of a call. These applications are still very immature, but their potential is great. DMG believes that real-time applications are going to play an essential role in the future of the speech analytics market.
Additionally, when a predictive analytics engine combines post-call and real-time speech analytics as one of the feeds to help identify the next-best action, it can change the dynamics of servicing. This is the future direction of these applications. Speech analytics is a proven and powerful solution, and combined with other analytics capabilities, its contributions increase dramatically.