Subscribe to the InfoTech eNewsletter

infoTECH Feature

November 19, 2012

How Obama's Campaign Used Big Data to Win the Election

By TMCnet Special Guest
Winston Christie-Blick, Product Management, GoodData

How did Barack Obama overcome the specter of an ongoing recession to win a second term in office? Obama’s campaign embraced the power of big data.

In the days following Obama’s victory in key swing states, anonymous members of the president’s reelection team have unveiled some of the activities carried out in the under-lit, top-secret division of the campaign’s offices­ – ­dubbed 'The Cave.' What they revealed was an immense behind the scenes data effort that established new standards for the way political campaigns employ data to gain the competitive edge.

This election season, Obama’s analysts continued on from where the 2008 campaign team left off, implementing lessons learned four years ago from the start. They began by consolidating their existing data from multiple databases. The next step was to collect new data through ongoing experimentation, and use that data to inform their day-to-day decision-making. Such data driven learning through experimentation led the campaign to discover universal best practices for recruiting volunteers, attracting donors, and winning over voters across the country.

But perhaps most significantly, big data also allowed the campaign to get small – that is, to target and galvanize specific demographics within high-stakes regions in crucial swing states.

Consolidating the Data: Lessons Learned from Past Campaigns

Before defining the role of data analytics in this election cycle, it’s worth noting that both of Obama’s campaigns have openly emulated tactics first used by George W. Bush’s 2004 campaign, which had a similar focus on using personal data to target voters. Still, according to a New York Times article from March, even veterans of George W. Bush’s reelection effort agree that Obama’s team is now able to employ data on a far larger scale.

With the recent proliferation of social networks and cloud applications, they simply had far more online data at their disposal.

“What is new is the power of the Web, the sophistication of what you can do to target people on the Internet, which is 100 percent new and continues to evolve,” said Sara Taylor Fagen, a senior strategist in the 2004 Bush campaign, now a specialist in online advertising and analytics.

This time around, the Obama campaign committed to consolidation. In the first months of the reelection campaign, silo’d data was merged into a single system, combining information collected from pollsters, fundraisers, field workers, consumer databases and social media websites. This enabled different teams within the campaign to collaborate over shared data, which may previously have only been available to the team that collected it.

Most importantly, insights pulled from the data were now transmitted to any team that could best act on them.

Collecting New Data Through Ongoing Experimentation

By consolidating its databases and fostering collaboration, the Obama campaign laid the groundwork for a data-driven effort that encouraged ongoing experimentation. To discover whether e-mails sent by campaign manager Jim Messina or by Vice President Joe Biden attracted more in donations, large randomized test groups were sent an email version “from” either Messina or Biden. The donations submitted by each group were then diligently recorded.

In this particular instance, Messina significantly outperformed Biden, but both were routinely trumped by Michelle Obama. Although one might suppose this sort of experimentation could consume significant time and energy, the difference between the outcomes of various approaches was hardly on the magnitude of dimes and pennies. Campaign data shows that top-performing e-mails, of the kind sent during the experiments outlined above, often raised 10 times as much money as their underperforming counterparts.

Experimentation led to other breakthroughs as well. By running various types of online campaign ads and analyzing the success of each, analysts were able to hone in on ad layouts that generated the most interest. One instance of such data-driven learning, noted by The Economist, was the discovery that campaign ads with the somewhat overzealous Sign Up Now buttons received far fewer hits than those featuring buttons compelling the viewer to Learn More.

Likewise, the campaign ran trials of new programs on small test groups before implementing them on a larger scale. This approach was carried out to evaluate the Quick Donation program upon its inception in Chicago. The new program was designed to simplify the donation process by allowing rapid donations via text and Web, and by saving donors' credit card information to facilitate future giving.

When it was discovered that donors in the program gave about four times as much as others, it was expanded and incentivized nationwide.

Perhaps most noteworthy of all, Obama’s campaign combined massive streams of swing state polling data, allowing them to check in on demographic trends within specific regions. So when polls started to slip after the first debate, it was clear which voters were changing sides. Campaign leaders knew exactly who to target. Such revelations directly impacted the way resources were allocated in the final weeks of the campaign, culminating in Obama’s reelection.

Getting Small with Big Data

This election cycle, big data ultimately allowed Obama's team to get small. By amassing information about different segments of the electorate, the campaign was able to target demographics of interest and custom fit its message to any audience. At times, the data supported turning to unconventional methods to connect with certain high-stakes populations. In the case of reaching out to Miami-Dade women under the age of 35, that meant abandoning traditional news programming ad slots in favor of those during shows like Sons of Anarchy and Don’t Trust the B in Apartment 23.

In this way, analysts in The Cave helped optimize the allocation of resources by ensuring investments made – in TV ads, phone calls, door-to-door canvassing, direct mailings, and social media – would be sure to have the greatest impact.

This commitment to “mass customization” helped the Obama campaign raise nearly $1 billion.

While it’s clear that the impact of big data on Obama’s reelection will not be lost on future campaigns, it’ll be interesting to see how others seek to replicate and augment the Obama team’s analytical tactics – both in the political sphere and in the private sector. In business, as in politics, we have moved beyond the era when competitive advantage can be derived from hunches, gut feelings, and appeals to what’s worked well in the past.

Across the board, in any field where competition is fierce, mushy pseudoscience of high-stakes management has been superseded. Its vanquisher and successor is data-driven learning – an iterative process of answering questions through experimentation and data analysis, only to uncover new questions of interest along the way.

Ongoing data driven learning informs the way we make sound business decisions on the macro scale. It also enables us to focus in on small areas of concern that might have gone unnoticed otherwise. With the right tools, the bigger the data, the “smaller” you can get. As the election news coverage fades into the distance, let’s not forget that the tools for gaining precious insights from your data aren’t restricted to The Cave. They’re just a click away in the cloud.

Want to learn more about the latest in communications and technology? Then be sure to attend ITEXPO Miami 2013, Jan 29- Feb. 1 in Miami, Florida.  Stay in touch with everything happening at ITEXPO (News - Alert). Follow us on Twitter.

Edited by Braden Becker

Subscribe to InfoTECH Spotlight eNews

InfoTECH Spotlight eNews delivers the latest news impacting technology in the IT industry each week. Sign up to receive FREE breaking news today!
FREE eNewsletter

infoTECH Whitepapers