The time had come: ‘Show me the value of big data’ was the message from our COO to the information technology and business unit teams. We welcomed the challenge. In May 2015, after a couple of years of conversations, demos and analysis, we were solidifying our understanding of our big data and analytics platform and its capabilities and wanted to test some of our ideas.
We began to challenge how we think about our business and approach our IT projects. Our organizational structure reflects the traditional core business units of an investor-owned, regulated electric utility: Generation, Transmission, Distribution, Customer Service, Commercial Operations and Shared Services. Our IT service model reflects this structure and we seldom have initiatives that benefit from cross-business unit collaboration – or so we thought. Our search for value through the application of advanced analytics has inspired common cause and collaboration that benefits the entire enterprise.
Now, with the power of technology’s latest capabilities, it is difficult to identify solutions that do NOT require us to collaborate across our traditional functional boundaries. We have implemented analytics projects aimed at improving the customer experience, optimizing our workforce and improving the reliability of the grid. The insights and corresponding calls to action often bring teams together and serve to connect the dots and broaden our thinking. We started with small and focused data sets and found that even with this pilot approach, unexpected insights emerged.
Segmenting Our Customers
Our Customer Service organization sponsored an initiative to build a comprehensive customer view to support AEP’s goal to improve our customers’ experience in doing business with us. We had several hypotheses that all required testing to improve our understanding of our customers, including proactive communications, improved call center services and performance, and broader adoption of available payment, savings and energy management programs.
We started the project with a customer segmentation analysis that – we soon realized – could not be accomplished without acquiring and integrating a broad set of information about our customers from many sources, internal and external. We included customer information such as usage and payment history, web site visits, mobile alert history, call center activity, program participation information, outage data and demographic information. The results were eye-opening – and led to conversations about how the segmentation and customer information could be used to drive additional analytics in several areas, including the following:
• Corporate Communications – our Corporate Communications team is interested in leveraging the segmentation results and additional customer information to drive improved customer web experiences and better target marketing and advertising messages. These practices might be familiar to the typical consumer products company, but they are new disciplines for a 100+ year-old regulated company.
• Operating Companies – our Operating Companies, which serve 5.4 million customers in 11 states, are excited to get a more comprehensive view of customers to help them design better products, services and outreach programs.
• Regulatory – The Regulatory group wants to leverage the information about our customers to drive meaningful conversations about how we can balance our capital investments to serve the objectives of the communities we serve.
• Program Managers – after sharing the segmentation results with our customer program managers, we found that the addition of propensity models and scores for particular programs will improve targeting.
• Call Center – we know that some customers will call repeatedly to find an agent that will give them a more satisfactory answer. We were able to identify two distinct patterns – customers who call multiple times in one billing cycle, and those who regularly call in the early or later stages of billing cycles over several months. We found that the majority of these calls were coming from one of our customer segments. We plan additional analysis of call details to implement solutions that will reduce those calls. We were also surprised to learn that, over a three year time frame, 73 percent of our customers had called our call center at least once. This revealed an untapped opportunity to market programs to customers with a high propensity score.
Improving Work Efficiency
Our Distribution Services organization sponsored a project to analyze the mix of employee and contractor crews that perform their work – installing and maintaining poles, power lines, transformers and other equipment – to determine if they could better optimize staffing for different types of projects. We leveraged three years of historical information and developed a forecast model that gives us a high degree of confidence in the most economic staffing mix by project type, allowing us to either reduce costs or increase productivity.
As we share the results across the enterprise, many other groups are now very interested in the model and results, including our Transmission and Generation organizations, which also leverage a lot of contractors. Our supply chain organization hopes to use the model to forecast the need for construction and maintenance materials with longer lead times, which could lead to improved pricing and lower inventory carrying costs. We see savings and efficiency opportunities worth millions of dollars across the company.
Our early investments in analytics to improve the reliability of the grid have grown organically within our Generation, Transmission, and Distribution organizations. We have initiatives in all three groups that are leveraging sensor and machine data (IoT) to identify patterns that can predict asset failures, possible tampering or incorrect installations or configurations. These teams have begun to share their experiences and we expect the algorithms and analytics will benefit everyone.
The Tip of the Iceberg
Our approach to analytics initiatives reflects a common cause spirit. Several years ago, we adopted agile development methodologies, allowing us to bring together cross-functional teams and focus on delivering value incrementally. We did not anticipate the need to fundamentally change how we think about data. Many activities that we previously considered useful but not necessary for success, such as maintaining a data dictionary, starting with logical data models or analyzing data lineage, we now understand are essential practices for successful analytics programs. With these tools and techniques, our IT architects and developers can accelerate our data science efforts by enabling analytic modeling and data modeling in parallel rather than delaying analysis in the pursuit of perfect data. The result is a stronger culture of collaborative, iterative development between IT and our business partners.
We are just beginning our journey with advanced analytics. We continue to adapt and adjust our old ways of thinking and working as we seek insights that will lead to actions to improve safety, efficiency, and lower costs. As AEP transforms to better meet the needs of our customers, taking full advantage of the value of data, analytics tools and techniques will help drive our success.