Call centers are oftentimes the most immediate touch point of businesses with their customers and the quality of call center service has dramatic impact on customer satisfaction. Operating large call centers is complex and expensive, and it requires a large number of employees, processes and technological solutions. Intuit operates over 30 call centers across eight locations, with thousands of agents, millions of calls and tens of millions of call minutes per year.

Improving the efficiency of our customer success agents and supervisors and introducing process automation enables increasing customer satisfaction with reducing call center costs. This large volume of calls results in a very big textual dataset. In this talk we will describe how we are using modern NLP methods to increase our call center efficiency, enabling better decision making and saving agent’s time. We will discuss the unique challenges in working with conversational data as well as the specific flavor that classic NLP offers the call center domain.