National Journal

If you're an average observer of presidential politics, the first interaction you have with a candidate likely is online. But did you know that when you look at a candidate's website, someone might be looking back at you?

Since the 2008 presidential campaign, digital operatives have used a process called A/B testing, or optimization, to target people who visit a campaign's website or mobile app. For political campaigns, that means tailoring messages to web or mobile visitors in a way that will entice them to vote, volunteer, or donate. And as wonky as it may seem, in the Internet age of political campaigning, you can't have good grassroots organization without good A/B testing software.

Optimizely, along with other software companies such as Adobe and Omniture, give organizations the tools to measure their interactions with users, and leverage those interactions to their benefit. But Optimizely has cornered the market on A/B testing in the realm of political campaigning. Using Optimizely's software, campaign operatives can tweak their candidate's website or app, tailor it to the user's personal data, and measure the "lift"—the difference between the "control" and the tweaked user experience.

Optimizely was founded by two former Google product managers in 2010. Dan Siroker, the company's cofounder and CEO, got the idea for it while working on President Obama's campaign in 2008. Today, Optimizely is the most adopted A/B testing platform in the world, according to Siroker, with more than 7,000 customers around the world.

In November 2007, Siroker listened to Barack Obama give a talk at Google. The then-presidential candidate talked about how he wanted to bring Google's data-driven strategies to campaigning, to harness the power of information for his own political engine. Two weeks later, Siroker quit his job and flew to Chicago to work at Obama's campaign headquarters.

Eventually, Siroker became the campaign's director of analytics, tasked with figuring out how Team Obama should use data to better draw in supporters using A/B testing. He quickly grew frustrated using the existing testing and optimization platforms, so, like any good Silicon Valley veteran, he decided to build his own.

The program Siroker helped build measured whether, by fluctuating different aspects of the Obama campaign's web presence, they could get more people to sign up for the campaign's email list, register to vote, volunteer, and donate.

"I think probably the most high-profile impact was on donations," Siroker says. "A/B testing directly led to tens of millions of dollars in incremental donations that the campaign would not have raised otherwise in 2008."

That year, Siroker conducted an experiment with a variety of splash pages—a page that shows up before navigating to the homepage—down to the text of the email sign-up button. The campaign tested four buttons: "Join us now," "Learn more," "Sign up now," and simply "Sign up." They also tested six media: three photos and three videos.

"It's the scientific method, but for the masses," Siroker said.

The winning splash page featured a black-and-white photo of the Obama family accompanied by the "Learn more" button, which had a 40.6 percent improvement over the original homepage's email sign-up rate.

It may seem trivial to focus on which slight variation led to the most positive user feedback, but the results speak for themselves.

"Roughly 10 million people signed up on the splash page during the campaign. If we hadn't run this experiment and just stuck with the original page, that number would be closer to 7,120,000 sign-ups. That's a difference of 2,880,000 email addresses," Siroker wrote in 2010. "The additional 2,880,000 email addresses on our email list translated into an additional $60 million in donations."

The Obama campaign enlisted Siroker's services again in 2012. So did Mitt Romney's campaign. Zac Moffatt, who was Romney's 2012 digital director, had his doubts about signing up for a service created by a former Obama campaign operative. Moffatt once said that signing up for Optimizely was the hardest decision he had to make on the campaign, but ultimately it paid off.

"We had massive trepidation because this guy had been the analytics director for Obama," Moffatt says. "But on the flip side, we believe in the market, and his product was the better fit."

While Romney didn't win, Moffatt was happy with the results his campaign saw using Optimizely's A/B testing. Similar to what Siroker did with Obama's splash page in 2008, Moffatt used Optimizely to test five splash pages that went to Romney's campaign homepage. With the optimized homepage, the campaign saw a 630 percent increase in email sign-ups.

"We have the statistics behind the scenes to show you that the difference would have been horrible," Siroker said. "One of the beauties of A/B testing is you know the alternative."

Siroker wouldn't confirm whether any 2016 presidential candidates have sought out his firm's services, but seemed confident that campaigns would.

This year, Optimizely is venturing into new areas of experimentation. The company recently launched a product called Personalization, which will allow campaigns to mine user data, trawling visitors' location data and other personal information to decide the best message to market to them. The more information that campaigns have on visitors, the more effectively they can target them. The most common piece of information that campaigns look at is user location. If you're visiting a Democratic campaign's website from California, the copy you see (and the emails you receive) are likely much different than what a visitor from Iowa sees.

Siroker said he doesn't see campaigns' use of A/B testing as malicious in any way; just a creative application of the scientific method. Looking at it cynically, mining Internet users' data for political gain sounds vaguely Orwellian, where voters are turned into unwitting guinea pigs in a giant social experiment. Looking at it pragmatically, it's just smart business.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.