Conversation intelligence is software that turns every recorded customer call into text and checks it. It finds the moments that matter, scores every agent on every call, and guides the agent while the customer is still on the line. It is also called speech analytics or conversation analytics.
Why is it needed? A contact centre records every call, because the rules in many industries require it. But until recently, almost nobody listened to the recordings. A supervisor could score only two or three calls per agent a month. The rest were stored and never heard. So a company knew very little about what happened on most of its calls.
The software does four jobs. First, it converts each call into a written transcript, through speech recognition. It also marks which words the customer said and which the agent said. Second, it finds the moments that matter. These include a complaint, a customer threatening to close the account, or a missed disclosure. A disclosure is something the agent must tell the customer by law or by the regulator's rules, such as the charges on a loan. It also judges whether the customer sounds calm or angry, which is called sentiment analysis. Third, it scores every agent on every call, against the company's checklist. That job is described in automated quality assurance. Fourth, during a live call, it shows the agent the right answer on screen. That job is described in agent assist.
Consider a company that gives gold loans, with a contact centre that handles 60,000 calls a week. Each week, the software reads every call. Say it finds three patterns. In 2,100 calls, customers asked why the amount credited was lower than the loan amount. In 900 calls, agents did not tell customers the processing fee before the customer agreed to the loan. And in 300 calls, customers said they would move their gold to another lender. The compliance team retrains the agents who missed the fee disclosure. The product team rewrites the loan SMS, so that it shows the fee and the amount credited on separate lines. The retention team calls back the 300 unhappy customers.
So the job of the quality team changes. It no longer listens to a few calls. It decides what to do about the patterns that the software finds across all calls.
The software receives the recordings from the contact centre software. It sends agents' scores to workforce engagement management (WEM), and the themes in what customers say to customer experience management (CXM). The same idea, applied to sales calls, asks a different question: will this deal close? That is revenue intelligence. A business phone system's call transcripts and AI summaries write up a single call for the person on it. Conversation intelligence studies all the calls together, for the company.