Data Rather than Door Knocking to Acquire Energy Efficiency Customers

Share Button

To acquire energy efficiency customers, many utilities hire people to knock on doors, especially for programs like demand response, says Sid Siddhartha, founder and CEO of Innowatts, which provides data-driven energy management.

The company has developed a program that uses data instead.

The Sacramento Municipal Utility District has begun working with Innowatts to use the program to help identify energy efficiency customers and identify efficiency programs best suited to them.

At the core of the program is what Siddhartha calls machine-learning energy analytics.

That means the company’s systems identify customers and programs, and then adjust their predictions–or the systems “learn”–based on information gleaned from the customers.

“Our machine-learning enabled energy analytics programs will allow SMUD to know consumption patterns at individual customer levels,” he explains. “This allows SMUD to target specific programs to specific customers.”

The program would identify, for example, air conditioning upgrades or home weatherization programs.

“This will allow SMUD to reduce the amount of money upfront in acquiring these types of customers; the technology can help them in a scalable way,” he says.

Siddhartha says the approach slashes the cost of acquiring those customers by up to 90 percent (in the case of demand response customers).

Utilities generally begin their search for energy efficiency customers using marketing techniques: They look at customer demographics, for example, he says.

“Our is bottoms-up analytics. It creates individual load profiles in conjunction with data about property and layers it with socioeconomic data, creating individual product profiles for each customer,” he explains.

The machine learning comes in after the program has identified specific customers for specific programs. “The platform might identify you as a customer for an AC upgrade for your specific address, but we might find this isn’t a good candidate, then we give the data back to the model and the model learns from there,” Siddhartha explains. “Maybe the model missed certain parameters. It takes information from people and learns from that.”

“As we get the feedback from the campaign, the machine starts learning and new learnings are incorporated into the next round of the campaign,” he says.

In other words, the machine identifies customers and programs, with the goal of helping utilities save money. But, luckily, it still needs a little help from humans.

Follow on Twitter @EfficiencyMkts.

Share Button