The Calabrio Customer Connect event was held recently in San Antonio, Texas. Designed for Calabrio partners and customers, the event was distinguished by outstanding participation and insights. As a strategic partner, Voci was in attendance, as were fifty other Calabrio customers and partners. We are looking forward to the next C3 Conference at the Hyatt Regency in Minneapolis, MN, on October 25-28, 2020.
There were many insightful breakout sessions and workshops during the conference. I had the opportunity to speak at two of them, and wanted to share some of my thoughts as I did during the sessions.
We see two large, but complementary, forces that will drive the delivery of better, faster and more valuable analytics in the call center of the future. The first force is the continued improvements that we see in machine learning, including both developments in the academic world, and increased application of machine learning in industry. The second is the continuation of Moore’s law, which suggests that continued improvements in semiconductor and related technology will result in continued increases in computing power, without increasing the hardware footprint.
Together, these forces will drive a shift away from descriptive analytics (e.g., what is being done wrong or what is being done correctly, by who, when, etc.) to prescriptive and predictive analytics that not only capture the “what”, “who”, and “when” but provide insight into “why” and “how” to fix it. Continued improvements to hardware will enable smarter machines and simultaneously enable the delivery of these more valuable, more actionable insights in real-time to affect the outcome of a call.
So, for example, rather than knowing after a call is finished that the customer is likely to churn, and being able to plan for the next call, we will be able to know this — and know what to do about it — while the call is taking place.
Real-time speech analytics are currently being used in three applications. First, real-time analytics has been applied in highly regulated environments for some time – such as the collections industry – to ensure agents are using prescribed language in their client interactions, such as properly reading the “mini-miranda”, not using legally threatening language, etc. Combining these simple, but useful, descriptive analytics with real-time emotion information gives supervisors the opportunity to understand which calls may need escalation before it’s too late.
Second, by combining deep learning with real-time speech to text, and real-time measurement of emotion, significant strides are being made to improve the outcomes of outbound sales calls. These state-of-the-art systems are designed to raise the performance of less well-performing agents closer to that of the top performers, while enabling the best performers to perform even better. The results are astonishing – as much as a 20% improvement in closure rates.
Lastly, real-time analytics are being applied to improve the agent & client experience by automating routine back-office tasks that otherwise disrupt the conversation flow. This disruption results in longer calls, less happy customers, frustrated agents, and lower first call resolution scores.
Preparation for these exciting improvements in call center analytics needs to start now. Given the increasing reliance on machine learning to drive superior analytics, customers must locate a trusted partner with whom they are willing to share data. Machine learning works by finding patterns in massive amounts of data that can be delivered as actionable insights. As a result, the willingness and ability to share data is very important.
This trusted partner needs to have the ability to deliver real-time when you are ready. Proven ability is key to being able to go to real-time when the opportunity presents itself, rather than scrambling to scale up old technologies or build something new.