Trying to derive meaning from a spoken word conversation can be difficult. Body language, tone of voice, even word choice all can have a meaning that shades a conversation in fashions so subtle that it's easy to get two different meanings from two different targets.
When that falls to just a voice, as in a phone call, that can make things even tougher since so many cues are lost. VoiceBase's new predictive analytics tool may be in a position to help.
This tool, known as Insights, uses a combination of predictive analytics and machine learning systems to take large amounts of recorded conversation and analyze them, looking for patterns in the system to apply to future calls. The system will automatically score calls as well, looking for things that might indicate a first-time caller, a hot lead, or even a non-prospect. The analysis is set to work both ways as well, noting things like potential churn and even rude agent for support calls.
On the one hand, it might be a bad idea to let the machines determine just what qualifies as a rude agent; call center work can be difficult enough without letting the machines be yet another boss to obey. On the other hand, however, this has some real potential.
It would be easy to overestimate the capability of something like this. After all, we're talking about machines here, machines that can only understand human emotion based on what someone else said a certain emotion looks like, and how closely the newest thing approximates that. But even with a grain of salt, a business able to tell that a caller really does sound interested or hesitant can mean it can use a whole different sales strategy. That different sales strategy can mean the difference between sales gained and sales lost.
Using pre-tagged calls as the basis for teaching its machine learning systems, Insights works through huge amounts of data to handle complex matters. This is a development that has many firms excited. Telmetrics is poised to bring Insights to its own operations, calling the system a platform that “has the potential to unveil a deeper, much sought-after layer of speech analytics, that up until now has been inaccessible.”
Edited by Rory J. Thompson