Although it’s a slow evolution, enterprises are coming to realize that big data can be extremely valuable. In the contact center, knowledge about the customer and history of customer interactions can and should affect how agents engage customers. But big data typically requires big dollars and big disturbances to your existing systems. If you want to get a broad and actionable view of a customer, you might need to integrate with the CRM, demographic/psychographic databases, purchase history, payment history, and account transaction history, to name a few. Changes to your legacy systems of that magnitude could take years. It’s no wonder that the use of big data is evolving slowly.
But there is another form of valuable big data that does not require complex IT work yet provides a good picture of the customer state of mind: the very words being spoken during the customer interaction. Aspect has introduced a technology that permits real-time, dictionary-independent speech analysis of customer and agent dialogs. We have integrated this new technology with our existing Notify solution, which can initiate prescribed screen pops, emails and SMS messages, and that combination opens up a whole new world of possibilities for the contact center. Essentially, the system can “listen” for key topics (or lack thereof) and initiate real-time actions related to what has been spoken. The challenge is to know what to listen for, and those topics are created by mining the massive volumes of existing calls (aka big data) to determine what transpired in those conversations with the desired outcomes. Over a period of years, we have developed a library of business topics associated with each vertical market. These libraries save huge amounts of analyst time, because we have already created the topic models based on the successes of other clients in each vertical market.
Let’s take a very simple example of how to drive higher cross-sell with real-time speech analytics. Assume a broadband company CSR has been trained to cross-sell high speed internet whenever the customer asks to enable new VoIP service. If the real-time analytics solution detects the customer asking about upgrading to voice over IP by saying, “Voice over IP” or “VoIP” in connection with “upgrade”, the agent receives a screen pop reminding him or her to cross-sell high-speed internet. The real-time speech analytics then listens for the agent to make the offer. If the agent does not offer high-speed internet within a configurable period of time, say 20 seconds, a second reminder is popped to his or her screen and a coaching session may be automatically scheduled. In extreme circumstances where the cross-sell still has not been made or if the call becomes contentious, the supervisor can be notified via screen pop or email/SMS to a mobile device.
The result is that a much higher percentage of cross-sell offers will be made, because immediate action is being taken to ensure that the agent does what he or she is trained to do. Although this example is rudimentary, it can dramatically affect conversion rates, and real-time recognition of other topics can be even more powerful, driving legal compliance, loyalty enhancement, cost reduction and most other enterprise objectives. You will doubtless see more about real-time speech analytics in the coming months as more enterprises adopt it as a shortcut to value in a world of big data.