Monday, March 07, 2011

Automating the call scoring process with speech analytics

Speech analytics used to be a big, intimidating, expensive undertaking, offered only by the call recording industry giants.

Now we've seen 8 or more other call recording vendors release their version of speech analytics.

Almost all of them have decided to invest in a technology from the UK rather than developing their own product - including my company, CSI.

It's a good business decision -- why reinvent the wheel when an existing, proven technology is out there with OEM / white label availability. Saves time and resources.

Some of the call recording companies have decided to just go to war with the same out-of-the-box vanilla features which come with the API.

A few, like CSI, have fully integrated the speech engine within their own optimization suite.

Some have taken it a step further and created unique, differentiating features. One such feature is the automated scoring feature from Virtual Observer Call Insight.

As soon as a call is indexed by the phonetic speech engine, Call Insight scores the call based on scoring forms built by the customer.

10,000 calls recorded on tuesday? 10,000 calls are scored.

It's pretty dramatic when compared to a live supervisor, who can score, maybe, 30 calls a day. Of course, you'll still want to use a supervisor to evaluate your most important calls. The speech technology can also help you more efficiently identify which calls are those "most important calls".

The automated scoring process works best in a script adherence format, where questions can be answered with either a true/false or yes/no type of answer. Most existing scripts can be refashioned into workable scripts for Call Insight to do it's thing.

It's extremely powerful for telemarketers, who are typically script-driven to begin with, but as we're seeing with more and more different types of implementations, it can be applied in many different industries across many different types of contact centers.


Enhanced by Zemanta