It’s a low-key, black-and-white photo of the Obama family circa 2007, but it was the reason the future president raised an extra $60 million for his campaign.
The picture was part of an experiment run by the director of analytics for the former president’s campaign. Dan Siroker wanted to test what would best convince people to support Barack Obama, then trailing both Hillary Clinton and John Edwards in the Democratic presidential primaries.
The former Google project manager, hired by Obama in a typically forward-thinking move to embrace data analysis and experimentation, decided to see what resonated best with the public.
“We had several ideas on how to convince people to volunteer or register for the campaign,” Siroker told News.com.au. “Through A/B testing, we were able to discover what worked.”
The team tried showing different images and calls to action to different users who visited the website, and the results were surprising. Rather than responding most often to a “stern picture of the candidate,” people were most likely to click on the more personal, family photo — “a very unusual idea at the time.”
It worked far better than a bells-and-whistles video the team thought would be the biggest hit. Had they gone with the video, they would have made “a huge mistake” because it did worse than all of the images, Siroker says.
The 2,880,000 Americans who clicked on the image ended up donating an average of $21 over the campaign — a total of $60 million in donations.
Siroker ran a dozen more experiments that were vital to Obama’s success. “We looked at video versus images, the latter was much more effective. We ran several experiments around donations and found your highest previous contribution is very influential. If you donate $25 in the past, it’s best to ask for $25 again. If you donated $25,000, that’s potentially a big opportunity missed.”
The data guru has since co-founded Optimizely, which takes this lesson of experimentation and expands it. He is now in Australia sharing his knowledge with major companies including Atlassian and Fairfax Media, but he says there’s no one-size-fits-all rule.
“At its most basic level, it means coming up with a new idea and testing it out in real life,” he said. “One thing we found time and time again is there are very few universal truths.”
“The problems have gotten much more complex.”
“That’s what made Google so successful. If you don’t have experimentation, you’ve missed an opportunity.”
Siroker says it’s no longer about “HiPPO,” or “highest paid person’s opinion,” but about using ideas and knowledge from across a team.
Tech giant Amazon is also doing this brilliantly — providing a culture in which ideas are tested on a small scale and live or die depending on their success. This is how the company developed its recommendations pages, search results, design layout, Prime and Echo — “the kind of ideas that would have been killed in a meeting before they came to life. It’s also how the firm was able to easily ditch the unpopular Amazon Fire.”
“That’s why it’s so perfect,” he explains. “You capitalize on the windfalls and mitigate the losses.”