Computational modelling of speech data integration to assess interactions in B2B sales calls

The business sector now recognizes the value of Conversation Intelligence in understanding patterns, structures and insights of authentic conversation. Using machine learning methods, companies process massive amount of data about conversation content, vocal features and even speaker body gestures of spoken conversations. This study is a Work-in-Progress (WIP), aimed to modeling the dynamics between sales representatives and customers in business-to-business (B2B) sales calls, by relying solely on the acoustic signal. To this end, we analyze 358 sales calls at the Discovery phase. To model the conversations, we compute a basic set of acoustic features: Talk proportions, F0, intensity, harmonics-to-noise ratio (HNR), jitter, and shimmer. The plots of each acoustic feature reveal the interactions and common behavior across calls, on one hand, and within calls, on the other. The study demonstrates that using delta metrics to assess the interactions leads to new insights.