Conversation Analytics can extract nearly 50 individual metrics from phone calls.
It can determine if a caller was upset, if an agent was polite, and even if the caller purchased something. It can tell if the caller was a good lead, if they were price sensitive, and even if the caller was confused.
In this blog post we’re going to describe precisely how Conversation Analytics knows that a caller was confused.
We’re even going to provide some of the phrases and keywords that Conversation Analytics listens for to determine if the caller was confused. If Conversation Analytics hears these phrases it can say, with confidence, that the caller was confused.
Traditional call tracking only provides data about which referring sources generated calls. In short: it tells you what happens BEFORE the call. That’s useful, but it’s not enough. It’s actually relatively surface data. Imagine, for example, if a web analytics platform only described what happened before the visitor arrived on your site? That would be a crappy web analytics platform.
Conversation Analytics tells you what happened ON the phone call. That’s deeper analytics.
Conversation Analytics take call tracking to powerful new levels.
Conversation Analytics uses sophisticated speech recognition technology to analyze what happened on the phone calls. Convirza has then layered proprietary algorithms on top of the speech recognition technology that seek to answer specific questions about each call. These algorithms analyze specific words and phrases that appear in the call itself—words that the agent and the caller actually speak. Based on the words and phrases that Conversation Analytics ‘hears’ it can determine if a caller purchased, was a good lead, or even if a caller was confused.
Conversation Analytics expresses every element it tracks on a 0-100 scale. The higher the number, the more confident Conversation Analytics is that an event actually happened on the call. For example, if the indicator level for Caller Confusion was 98, that means Conversation Analytics is nearly certain that the caller was confused. If it was 55, it is still likely that the caller was confused, but it is also possible that the caller was not confused.
(Generally anything over 75ish is considered quite certain. You should feel comfortable automating Convirza Webhooks based on that 75 threshold).
There are over 400 specific phrases Conversation Analytics listens for to determine if a caller was confused. We’ll only list a few here. And keep in mind that all of these phrases need to be in the proper context for Conversation Analytics to conclude that the caller is confused.
– Better able to explain
– Can you explain again|could you explain again|how|it|once|one|more|that|this|what|why
– Can you just explain this
– Can you repeat|would you repeat|you repeat again
– Doesn’t|didn’t|make sense
– Don’t know what that means|not sure|can’t figure out|really confused about
– I’m confused|baflled|frustrated|annoyed
– Mind repeating|repeat again
As you can see, there are several–hundreds–of ways that a caller could indicate confusion on a call. And again, all of these phrases need to be used in context for Conversation Analytics to assign any value to them as an indicator of caller confusion.
This is cool.
What’s even cooler is all the things you can do with this data.