A Fortune 50 company sought to better understand the underlying reasons for customer service calls. To accomplish this, the company needed more insights from data stored across multiple, disparate data sources.
As summarized in Figure 4, Maana’s Knowledge Platform was deployed to identify the issues that prompted customer service requests. Maana crawled and indexed multiple data sources, such as global installed base data, turbine trip data, controller alarm data, parametric time series data and field service data from their ServiceNow system. The project involved over 600 million alarm values and approximately 1,200 trip events.
Natural language processing (NLP) was used to identify the reasons why people called customer support. Machine learning capabilities enabled subject-matter experts to gain a better understanding into the causes of turbine alarms. Specifically, Maana’s Knowledge Platform found and identified correlations between historical patterns in customer support calls and trip events. The correlations showed obvious but previously unverified relationships between trips and support calls for various generator models. Using these insights, the company can better understand the underlying triggers for customer service calls and determine which product lines and regions were incurring the most issues.