Conversation Analytics: Can Machines Read Between the Lines in Real-Time Strategic Conversations?

Strategic conversations involve one party with an informational advantage and another with an interest in the information. This paper proposes machine learning–based methods to quantify the evasiveness and incoherence of the more-informed party during real-time strategic conversations. To demonstrate the effectiveness of these methods in a real-world setting, we consider the question-and-answer sessions of earnings conference calls, during which managers face scrutinizing questions from analysts. Being reluctant to disclose adverse information, managers may resort to evasive answers and sometimes respond less coherently than they otherwise would. Using data from the earnings calls of S&P 500 companies from 2006 to 2018, we show that the proposed measures predict worse next-quarter earnings. The stock market also perceives incoherence as a negative signal. This paper contributes methodologically to business analytics by developing machine learning methods to extract behavioral cues from real-time strategic conversations. We believe the wide adoption of these tools can increase the efficiency of various markets and institutions where real-time strategic conversations routinely occur, which ultimately benefits business and society.