A new technology launched today is capable of analysing and interpreting human to human spoken conversations. It has been developed quietly over the last two and a half years by San Francisco-based Utopy.
The technology forms the basis of Utopy's Speechminer software programmes which are already being used by financial institutions and Fortune 200 companies to analyse calls. According to Utopy's president and CEO, Roy Twersky, "With the first ever successful customer deployment of speech mining technology, Utopy's SpeechMiner has delivered on the promise that a software solution would understand the spoken dialogue of telephone conversations. It opens completely new possibilities to the enterprise by turning voice interactions into business intelligence and actionable information that has never been available before. SpeechMiner solutions demonstrate significant and immediate return on investments to Fortune 1000 companies by helping them increase sales conversion rates, automate contact centre processes, reduce customer churn rates, and increase revenue by understanding their customers' needs."
Most contact centres record conversations with customers and file them away, only to be retrieved if a specific issue arises. SpeechMiner uses a comprehensive set of algorithms to intelligently break down each telephone conversation into multiple linguistic and non-linguistic events, including agent and customer words, phrases or emotions. Once the calls are broken down into specific events, its classification algorithms analyse the events and categorise each according to a set of business-specific rules that would be of interest to the organization.
Macro and micro levels
Contact centres can quickly identify trends, trouble spots and opportunities at both the macro and micro level. At the macro level, for example, managers can monitor overall levels of customer satisfaction and success of sales and marketing campaigns for entire groups of customers. At the micro level, managers can identify specific customers who are ripe for further sales, monitor the quality of individual agents, ensuring that they keep to policies and scripts, and flag dissatisfied customers and customers at risk of churning.
More accurate than human
The system, which can be up and running within two weeks or less, has greater than 99% accuracy. In a recent test that pitted SpeechMiner against a human contact centre call reviewer, SpeechMiner was 10 times more accurate in identifying the conversation event rules established by the contact centre manager.