DTE Energy Case Study

Increasing efficiency through greater insight

DTE Energy is a diversified energy company that provides electric and gas services to more than three million residential, business and industrial customers in Michigan. The economic conditions in DTE's market are difficult: Michigan has the highest rates of unemployment, bankruptcy and foreclosure in the nation. As a result, DTE wanted to identify ways it could work more efficiently, without sacrificing customer satisfaction.

DTE chose Intelligent Results PREDIGY platform to help them gain insight and increase efficiency, while continuing to serve their customer base. We recently spoke with Ozgur Tuzcu, a Principal Analyst in the Customer Service Organization at DTE. Ozgur talked with us about how DTE is currently using the PREDIGY platform and ideas he has for the future.

Choosing Intelligent Results PREDIGY platform

Due to the local economic conditions, DTE is faced with a large volume of accounts in collection, and they were looking for a way to better understand those portfolios. In the past, they had segmented accounts using factors such as balance, age of debt, and account history, but Ozgur says, "We wanted to take a more scientific approach in predicting customer behavior."

DTE evaluated the products on the market and chose the PREDIGY platform. They particularly liked the fact that PREDIGY enables them to create models based on their own data. With PREDIGY, they can use the champion/challenger method to statistically compare new strategies with "business as usual." Ozgur says, "Champion/challenger is extremely easy with PREDIGY tools, and it's one of the reasons we went with Intelligent Results." In addition, the reporting makes it easy for them to understand the distribution of a factor within a certain population. In short, Ozgur says, "we went with Intelligent Results because of its capabilities and because it makes life easier."

Less Effort, Better Results

As DTE analyzed its accounts in collection, one goal was increased efficiency. That is, they wanted to maintain the amount collected while reducing operating costs-and to document lessons learned for future use. They analyzed their accounts and were able to prioritize efforts based on a charge-off model score. Using champion/challenger strategies for account groups in various stages of collection, they determined that most early-stage accounts provided better results when contacted less frequently. That is, people were more likely to pay if DTE contacted them less frequently. The exception to this was low-scoring accounts with low balances.

While this result was not entirely a surprise to Ozgur, he says that "the differences were much greater than expected. It's pretty nifty when that happens—we bring in more money with less expense. In part, it's a process of getting to know our customers better."

Better Service at a Lower Cost

Beyond collections, Ozgur's group is finding other ways that PREDIGY can help DTE increase efficiency while maintaining customer service. For example, they have recently used PREDIGY to do research in the area of service disconnects.

A service disconnect is when the company discontinues a customer's service due to unpaid bills. Service disconnects are expensive for the company and disruptive for customers, who often pay their bills and have the service reconnected within days. DTE was looking at ways to improve the process for both the company and its customers.

Ozgur told us that there are two problem areas related to service disconnects. The first is that meters are sometimes inaccessible. There might be a locked gate, or a big dog, or something else that makes it impossible to get to the meter. A field collector who is sent to the site makes a note of the situation and moves on. If DTE knew in advance that the meter was inaccessible, then they wouldn't incur the cost of sending someone to the location.

The second problem area is that people often pay their balances and have their service reconnected within days. This is expensive because DTE has to send two people out to the location: one to disconnect the service, and another to reconnect it several days later. It would be better for DTE—and the customers—to have customers pay their past-due amounts before service is disconnected.

DTE used PREDIGY to analyze these two areas: whether the customer is likely to be there for the disconnect and whether they are likely to reconnect within 7 days. They found that the customers who are most likely to be there are not likely to reconnect within 7 days. And, furthermore, these accounts are the ones that are most likely to be written off eventually. While DTE is still in the process of changing its operations to reflect these findings, Ozgur sees that "there is potentially a big gain in this area."

Future Plans: Finding More Ways to Work Efficiently

We asked Ozgur about future plans, and he said that there are three models that they would like to build. One model will be designed to predict the probability of a recovery correctly. This would enable DTE to work more effectively both in-house and with outside agencies. A second is a phone collection model, designed to predict whether a customer would make a payment or a promise to pay if DTE called. Again, Ozgur sees potentially big gains in terms of efficiency.

Ozgur also would like to use analytics to assess cases of possible theft. DTE has a system-generated list of events that is used by the people in the theft office. This list includes tens of thousands of accounts. Some of these are billing errors and are handled accordingly, while others are targeted for investigation. Ozgur hopes that PREDIGY can help identify the probability of a theft, so that the people in the theft office could focus their efforts accordingly. This would enable the theft office to work much more efficiently.

Analytics in the Enterprise

Beyond these areas, Ozgur says that he has had discussions with other people at DTE who want to use predictive analytics. For example, analytics have sparked interest among the people responsible for electrical generation who would like to be able to predict equipment failures. When equipment does fail, it can damage other equipment upstream or downstream. If they can predict likely locations of failure, they can shut the equipment down properly and avoid the damage caused with failure.

There is also enterprise analytics. Ozgur says, "really, analytics can be useful in any place where you would like to be able to make predictions and bring down costs. I see big potential in that. In our case, we can decide whether to keep our effort the same and reduce costs, or make decisions that allow us to manage our business in entirely new ways. It becomes a business choice at that point. A fact-based choice. And that is useful in any business."

About DTE Energy Co.

DTE Energy Co. is a diversified energy company that provides electric and gas services to more than three million residential, business and industrial customers in Michigan. Ozgur Tuzcu is a Principal Analyst at DTE Energy. His responsibilities include implementing predictive analytics, developing and implementing scoring models, and integrating the resulting models with operations. For more information, visit www.dteenergy.com.