Chirag Shah, Vertex 2017-08-31 04:52:29
Utilities now have more customer data access than ever. This information is more than just data from advanced metering infrastructure (AMI) or from customer information systems (CIS); there’s a wealth of third-party information that can reveal customers’ likes, dislikes and behavioral propensities. At the same time, today’s consumers expect businesses to know and cater to them as individuals. A quick Google search on customer service best practices produces terms such as personalization and engagement. Wouldn’t it be great if you could use that mountain of data to better understand your customers and give them the personal attention they demand? Even better, what if you could use it to increase operational efficiencies and optimize spending at the same time? Predictive analytics can help utilities do this across the entire customer life cycle. The importance of predictive analytics Most utilities extract performance and usage reports from collected data. While such descriptive analytics are important, they look at the past to describe what happened. By connecting customer and usage data with predictive analytics (PA), a utility can begin to uncover customer propensities and tailor their engagement. PA leverages machine-learning algorithms for data mining to identify key drivers of customer behavior. The predictive nature isn’t based solely on a direct customer observation or experience. By supplementing historical customer behaviors with external data sources, PA predicts a customer’s behavior based on similar audiences. These predictions give you actionable insights you can use to optimize everything from marketing to operations. Here are four examples of how a utility can leverage PA across the customer life cycle: 1 Attract profitable customers Deregulation is coming to utilities, which means consumers can select from a list of providers. That means utilities must compete for their business. While you want to grow your customer base, you also want to spend your marketing budget wisely. Using a statistical technique known as logistic regression, which studies the association between a binary dependent variable and a set of independent variables, PA lets you identify prospective customers who are more likely to be profitable and less costly to serve. It can: • Identify prospects with traits and interests similar to those who previously responded to your messages. These might be demographics, buying habits or interests expressed online. This insight also lets you tailor the message for that prospect — something customers expect. • Prioritize prospects with higher lifetime value. • Deprioritize prospects with a history of overusing support services. The retail services division of a Midwest gas utility used predictive analytics to score 2.5 million potential acquisition targets and identified the top 20 percent of prospects for targeted acquisition campaigns. Three new acquisition campaigns were evaluated using a champion-challenger model. A champion process is the current best campaign, while a challenger is a new campaign driven by predictive analytics. The campaigns using the challenger model to identify prospects were between 130–300 percent more effective. By identifying the key predictive variables, the utility could refine marketing campaigns to resonate with target prospects, based on the customer segmentation. 2 Anticipate a customer’s likelihood to pay Customers fall behind on payments every day. For some, it’s accidental and quickly remedied. For others, it’s a sign of ongoing trouble. When a customer goes into arrears, what does your company do? Do you send a past-due notice? Or do you start the disconnect paperwork right away? How can you be sure which action is appropriate? The last, most expensive action you want to take is to disconnect service when a proactive reminder would have sufficed. By using PA to assess the likelihood of each customer paying on time, a utility can determine whether a gentle reminder or a sharp disconnection notice is appropriate. A Midwestern sewer utility reduced the number of customers entering arrears by 25 percent through proactive, differentiated credit and collections treatment of customers, based on their predicted likelihood to pay back debt. Customers were segmented using propensity scores calculated by the PA model, which is based on behaviors exhibited by similar customers in the past. Different strategies were applied to drive collections based on risk. The utility achieved a 30 percent increase in collections on delinquent accounts, without a corresponding increase in cost to collect. PA allows the utility to continuously tweak strategies based on customer risk segmentation to achieve incremental improvements over time. 3 Meet regulatory targets and increase your share of wallet We’ve seen that PA can identify a prospective customer’s likes and dislikes to help you attract more profitable customers. You can also use a similar strategy to persuade existing customers to take the actions you want. Some utilities have regulatory mandates with customer programs (e.g., energy efficiency). To achieve those mandates, you need customers to enroll in specialty programs. As with any marketing plan, you want the biggest impact for the smallest cost. Instead of blanket mailings, PA lets you target customers most likely to respond to a program, regardless of its nature. PA can also help increase your revenue by targeting customers who are likely to add-on services or respond to upsells — and those who are not. A Northwestern electric utility partnered with Vertex to leverage PA to build propensity models for eight different customer programs, and to identify which existing customers should be targeted with what programs. In pilot campaigns using the propensity scores, the utility achieved a 70 percent increase in program participation using the same marketing budget. In addition to the utility direct marketing campaign leveraging propensity models, a retailer store partnership was developed using predictive customer characteristics from the propensity model. The utility identified six zip codes for retailer store-focused campaigns and four-to-six retailers (Walmart, Home Depot, etc.) for trade allies. The propensity model identified predictive demographic characteristics of prospects, and these characteristics were used to identify the target zip codes, as well as retail stores where promotions could be launched. 4 Retain the customers you really want Finally, once you’ve attracted high-quality, low-cost customers, you want to keep them. Sadly, despite your best efforts, you can’t make all customers happy. While losing a customer is painful, losing a good customer is worse. PA can identify those customers with a high propensity to leave. Utilities can proactively target customers with a high propensity to churn and optimize incentives offered based on the projected customer lifetime value. A Southeastern deregulated gas utility developed customer treatment strategies based on PA scores and lifetime value. The utility offered a range of incentives from discounted rates and frequent flyer points, to gift cards for individual customers. With this approach, the utility could reduce customer churn and increase market share. Meeting customer expectations for a personalized experience While it may appear that the primary benefit of PA is to a utility’s bottom line, an equally important benefit of PA is that it gives customers the level of personalization and engagement they want and expect. You can tailor your message to attract the right customers, and know how to treat customers behind on their payment. You can recommend programs you know they’ll appreciate. Also, by knowing who’s likely to be a loyal customer, you can provide VIP treatment should they ever consider leaving. PA is a win for both utilities and their customers. It lets your company optimize spending to: • attract profitable customers, • increase enrollments in beneficial or mandatory programs, and • increase operational efficiency and waste reduction. A Southeastern water utility used PA to identify abnormal water consumption, and initiated a program that alerted customers of potential high use before they incurred large bills. The customers were extremely pleased with the new program to proactively notify them of a problem before it got too big. The utility could save on an important natural resource, while positively impacting customer satisfaction. CHIRAG SHAH is director of analytics and consulting for Vertex, provider of VertexOne cloud-based customer experience solutions for North American utilities.
Published by AMG. View All Articles.