Several weeks ago we released Conversation Analytics, a tool that tells businesses what actually happened on the phone call.
It analyzes the words, phrases and other cues spoken on the call to determine if the caller was a good lead, a sales ready caller, and if they ended up buying something.
Pretty cool stuff.
Conversation Analytics, however, doesn’t only track the customer side of the conversation. It also listens to the agent side of the conversation to determine how well the agent performed on the call.
Conversation Analytics can measure an agent’s sales performance, whether or not an agent was empathetic, and even whether or not an agent was polite.
This blog post will discuss how Conversation Analytics knows an agent was polite. In other words, what does Conversation Analytics listen for to produce confidence indicators for agent politeness?
Note: When we say ‘agent’ we don’t just mean a call center agent. An agent is anyone answering a phone at any business. Most of our clients, in fact, are not in a call center environment. Instead, they’re part of an inside sales team or answering phone calls in a retail environment at dentist’s office, tire store, or plumbing business.
To produce a confidence level (a number from 0-100) for Agent Politeness, Conversation Analytics essentially listens to the call content and tries to answer one question: was the agent polite?
It looks for changes in voice, rate of speech, lack of voice tremors, lack of agitation, and specific phrases and words. All of these things are ‘cues’ that Conversation Analytics uses determine if an agent was polite.
Conversation Analytics crunches all these numbers, adds up all this data, and then decides how confident it feels that the agent was polite. For example, if the agent politeness indicator is 78, Conversation Analytics feels quite confident that the agent was polite. However, if the agent politeness indicator is only 25, that means Conversation Analytics is not confident that the agent was polite.
As we mentioned above, Conversation Analytics analyzes various elements of the call, everything from voice tremors and volume to pitch and percent silence. But the most important thing Conversation Analytics listens for is specific phrases and words used in context with other specific phrases and words.
There are over 700 phrases for which Conversation Analytics listens to determine if an agent was polite. Below, we have only listed a few of the 700 phrases we look for.
– beg your pardon|pardon me
– May I please|help|serve|assist|monitor|bye
– Sir/Ma’am|alright|yes|no|please|thank you|repeat|will|name|yes|yeah
– Anything|something|any|I can help with
– Glad\happy to help|assist|some help|
– May I help you|ya
– Have a great day|awesome day|great day|good day|nice day|lovely day|wonderful day
– I’m sorry|apologize
When Conversation Analytics hears these phrases in the context of everything else on the call, it can determine with almost certain accuracy is the agent was polite or not.
In future blogs we’re going to discuss how to use the Agent Politeness indicator to make more money and hold employees accountable.