A major insurance carrier desired to improve close rates for call center agents fielding incoming prospect calls driven by various marketing campaigns.
While some agents were more successful than others, a clearly identifiable, scalable approach to improve all agent performance was difficult to identify.
Using natural language processing (identifying speech patterns from speech-to-text and machine learning across nearly 100 intra-call behaviors), segments of agent conversations were identified and tagged that either contributed to closing a sale, or detracted from the desired outcome. Several months of calls were analyzed for more than 25,000 calls (approx. 4,000 hours of talk time) per day.
As a result of this approach, sales within this call center increased by over $1.25 million in the first 12 months - a 30% increase over previous results.