Introduction

Opower (a part of Oracle Energy and Water) works on climate change. Specifically, they use data and behavioral science to “influence utility customer action at scale.” Basically, they are trying to get utility customers to use less energy, and to use the energy that they do consume more efficiently.

It turns out that utility companies in the US have to comply with regulations that require them to reduce the amount of power that consumers use during peak hours–weekdays from about 4pm-9pm in the evening. They have their own internal motivations too: servicing peak usage is an expensive proposition. It can require utilities to build extra power plants–plants that sit idle for most of the day, then need to be brought online during peak hours—an expensive process. (And a dirty one too: starting and stopping a powerplant is particularly bad in terms of greenhouse gas emissions.)

So utilities–and society in general–stand to benefit from a reduction in peak hour energy usage. Seems simple, right?

The problem here is that consumers use energy in ways that are hard to change. They come home from work, turn on the AC, run the washer and dryer, and use the oven. It seems like an immovable fact of life. This is true even when consumers have good intentions and want to help.

Opower set out to solve this problem.

Motivating Behavior Change

Opower addressed this work by relying on a key piece of research from the seminal paper “The constructive, destructive, and reconstructive power of social norms” published in the Psychological Science journal. It turns out that when researchers looked at different ways to motivate people to use less energy, the most effective motivator wasn’t money saved. It wasn’t altruism either. It turns out—at least according to this study—that the best way to motivate people was to compare them to their neighbors. If you tell people, “last month, you used 5% more energy than your peers,” that gets their attention and can motivate behavior change.

Impact, Outcomes, Output

I want to stop for a moment here to talk about behavior change. Longtime readers will know that this is a favorite topic for me. I write often about the Logic Model framework, which, simplified, can be seen in the diagram below:

In our story so far, we’ve talked about the Impact that Opower is trying to create. Opower wants to reduce greenhouse gas emissions. This is an example of a complex problem, and has many components that contribute to it. One of those components is consumer behavior. We’ve also talked about the Outcome that Opower wants to see here—they want consumers to reduce energy usage during peak hours. That’s how this model works: outcomes are valuable changes in human behaviors, ones that make a positive contribution in terms of the impact we’re trying to create.

Overcoming Obstacles

So now, the next question is this: what can we do or make to achieve this outcome? In this case, what can we do to get consumers to reduce their energy use during peak hours? Opower had an idea for a product called Behavioral Load Shaping, an email outreach program that shared usage data in an effort to influence consumer behavior.

Opower doesn‘t work directly with energy consumers. Instead, they sell software products to utility companies. This means that while Opower can develop products and solutions, they are all white-labeled solutions that are delivered from the utility company. And this, it turns out, is a bit of a problem.

First of all, most people think about their utility bill online for for around 8 minutes every year, according to Accenture. In other words, this is a low-interest category. So when utility companies reach out, people are likely to ignore that contact.

On top of that, when people do pay attention, they’re suspicious. If you’re like me, when your utility company reaches out to you, you think, what do they want from me?

It turns out that your utility company wants you to use less power. Really.

How Do You Create An Outcome?

Opower used a rigorous, product-led approach to their work. Before launching anything, they would run rigorous customer discovery, talking to homeowners, utilities and testing prototypes before launching. So Opower tested a number of different tactics in their outreach, all based on the foundational research: consumers respond best to peer pressure.

Their first email looked like this:

In small-sample testing, they discovered that this email didn’t work because they realized that most homeowners don’t understand what a Time of Use rate plan actually means. This is pretty typical for a first prototype—often you don’t even get to measure the prototype in terms of outcomes. Instead you’re just making a basic evaluation for readability, comprehension, usability, etc.

So they iterated.

They tried a second design. It was better, but it still wasn’t good enough. In this case, “good enough” meant that they felt confident to actually send it to utility customers.

The third iteration tested well. The team ran small-sample tests: the people understood their time of use rate and reacted with “oh I should do my laundry after 9pm!” This reaction was exactly what the product team was looking for—one that predicted the outcome they were seeking.

Now, they were ready to start some larger-sample experiments, experiments that would generate some measurable outcomes. They were ready to send actual emails to actual utility customers.

Measuring the Results

One problem facing companies that want to be more outcome-centric is that they often don’t have the data to work with to measure their work. Opower though is a digitally-native company, built on top of data. Their service works by measuring consumer’s energy use through their smart meters and utility bills. It gathers usage data, and then uses that data to generate insight.

In this case, Opower was looking to measure two things. First, would customers open the emails that they sent? Then, would those customers actually reduce their peak-hour energy usage? While all modern companies have the ability to measure email engagement, this measurement required deep integration with utility companies. Utilities provided Opower with emails of customers on time of use rates and then Opower automatically enrolled them in the program.

The program was framed as a Coach so many customers saw it as a helpful tool from their utility. In addition, every email would have specific tips about the specific actions to take to reduce electricity usage during peak hours. Customers were then directed to the utility website for more details on their usage patterns or for enrollment in other energy savings programs. In the initial pilot, Opower was able to demonstrate significant results.

The initial pilot launched to 35,000 customers of Exelon utilities in Maryland – Baltimore Gas & Electric, Pepco and Delmarva and the results were strong. In terms of outcomes, Opower saw strong engagement with the emails they sent, as well as very positive qualitative feedback from customers. Those initial results drove adoption with utilities across the country and how Opower offers this service to over 1M homes in the US.

Customer Satisfaction outcome was met: utility customers are happy 🙂

But email opens and engagements are just a leading indicator. The bigger question: would that behavior lead to the big behavior change they were seeking, reduced peak-hour energy use? It turns out the answer is yes. Opower saw a huge win, reducing peak-hour energy use by significant measures.

Peak Demand Outcome was met: Randomized Control Trials show 14 MW of Peak Demand reduction for ~1M homes in Mid-West utility

The Power of Outcomes

Sometimes, we’re faced with what feels like an impossible challenge—an impact that seems completely out of reach, or beyond the scope of our work. Fight Climate Change! Reduce Greenhouse Gas Emissions! In your business, you might see things like, Increase Sales! or Increase Profit Margin!

This story shows one way to make progress on these kinds of problems. Break them down into outcomes.

• Ask the question, “when we’ve achieved this impact, what will people be doing differently?” For Opower, the answer was people shift major appliance usage to off peak hours and thereby reduce peak-hour energy usage.
• Then, ask the question again, with a slight twist: what will people be doing that predicts this outcome? In this case, the answer is, “they’ll open and engage with our emails on a regular basis.
• Then, as a final step, ask, What can we make or do to create this behavior? In this case, Opower developed their Behavioral Load Shaping product, and it’s working.