Jump to Any Race
Political Advertising|By Elizabeth Wilner, January 10, 2013

Big Data did a cannonball into the political pool in 2012. President Obama’s re-election campaign aggressively—and ultimately, highly successfully—fused its internal polling with a wealth of consumer research to more directly connect polling to advertising, field, and ultimately, turnout. Among the ripple effects: Obama’s targeting trumped Republicans’ spending; analytics-enhanced polling gained serious street cred in the world of public opinion research; and new opportunities and approaches now confront the curators of these troves of information, not just the newly celebrified “quants” who analyze them.

Take television advertising, for example—the single biggest expense of any campaign. Aggregated information about TV ad content and placement provided by my firm, Kantar Media CMAG, and information from typically Madison Avenue-focused audience measurement providers such as Nielsen (TV program ratings) and Rentrak (set-top box data) fueled Obama’s unprecedented ad-targeting apparatus (as well as Governor Mitt Romney’s leaner, shorter-lived one).

In CMAG’s case, we effectively hooked up a fire hose between our servers and the Democratic National Committee’s (as we also did for the Republican National Committee) and let our data sets pour into a new program called the Media Tracking System, which merged our intelligence with that of the campaign’s media buyers and pollsters to help them track which TV ads moved their targeted voters, and how.

“One of biggest advantages to merging paid media spending, creative data and polling in the same application is that for the first time on record, all campaign staff could look at the same data at the same time,” said Chauncey McLean, the DNC’s Chicago-based director of media tracking. “In 2008, paid media data lived on two pieces of butcher paper. Polling data lived in reams of banner books spread everywhere. Imagine trying to have a meeting about the state of the race in Cleveland” with people dialing in from different places looking at different data. “Our state leadership couldn't believe how cleanly MTS could deal with such complicated data sets.”

The other big benefit was data visualization. “We could compare our polling in any media market against the paid media spending in that market over time. Lining that up showed us at a glance just how stable the race was. This was helpful not just internally (we could calm staff in places where we were being outspent by showing that Republicans weren’t moving our numbers), but it was also helpful with press.” Communications staffers could illustrate quickly through graphics that the GOP’s spending advantage wasn’t translating into a lead.

Other programs developed by the Obama team merged different combinations of data sets, as documented by Sasha Issenberg in his excellent series in MIT’s Technology Review. The common denominator was always their internal polling.

The headline of 2012 for those who conduct, commission or just crave polls? The ongoing transformation of opinion research will gain momentum—unlike changes over the past decade that have been largely reactive to cultural shifts like exploding mobile phone use and regulatory developments like caller ID. Not only will the practice of polling increasingly become a hybrid of analytics plus live and online interviews and robocalls, but the future of polling promises to be less about gauging public opinion than about shaping it. It will be less about getting a report in the New York Times or on “NBC Nightly News” saying “Candidate X is up by 4 points” than it will be about moving Candidate X to a 4-point lead.

This isn’t necessarily a healthy development for democracy. By making polling more actionable, the embrace of Big Data may make presidential politics smaller. Campaigns will be able to more efficiently target those who may be persuadable and who—per the preexisting filter of the Electoral College—could influence an outcome, and overlook those who aren’t persuadable or can’t affect an outcome. Only 65 markets saw any broadcast TV advertising by the Obama campaign last year. Only 52, or less than 25 percent of all media markets, saw more than 2,000 occurrences of Obama ads (the campaign’s average per market was about 8,200).

The ideas candidates push may become narrower. Critics accused the President of playing “small ball” by declining to outline ambitious new policy goals for his second term and instead dwelling on past accomplishments like the Lilly Ledbetter Act that appeal to certain demographic segments. The Obama campaign sliced and diced the electorate in ways that looked a lot like, well, product marketing.

There’s a headline here, too, for Madison Avenue-oriented Big Data providers: look to the south. You have an opening to do more business in the politics “vertical,” or space, though that opportunity isn’t without hurdles.

We likely will go for the better part of a decade—until the next incumbent president seeks re-election in 2020—before we see another campaign that can afford to invest this kind of time and resources in analytics. A senior Obama official calls analytics their third greatest expense after advertising and field. The luxury of incumbency and lack of a primary challenger gave them time to develop their groundbreaking dashboard.

As Charlie Cook has suggested, the responsibility for maintaining this infrastructure may shift to non-candidate entities such as the 501(c)(4) being created by Obama’s campaign or the think tank-like organization being discussed by a circle of GOP quants and media buyers. Another option would be to house the operations with the parties, which are freer to coordinate with candidates. Either way, relying on just a few seven-figure political clients isn’t a sustainable business model for data providers.

Nor are many accustomed to delivering their data in ways that are easily aggregated for political purposes. The Obama campaign bought data in bulk and deployed an army of programming whizzes to break down the silos in which these data typically live and connect them in real time. Vendors who can help less-equipped political clients do the same will find greater success.

The 2012 campaign provided a nightly lesson for the consumer marketing world. The key difference between product marketing and electoral marketing, a/k/a politics, is market share. A brand of automobile can succeed with a relatively small share of the driving population, but a candidate can only win with the largest share of the voting population. The Obama campaign took the type of ad agency data used for years to position automaker clients against multiple competitors and leveraged it, in real time, to defeat just one.

Presidential politics is where much of the innovation happens in marketing. Indeed, you could argue that presidential elections are marketing’s de facto quadrennial trade show, when all the new and improved tactics are rolled out for the mother of all trial runs. True, those on the consumer end don’t face constitutionally mandated, winner-take-all competitions that drive them to regularly update how they do their work, but some folks just dropped a few hundred million dollars on showing them how they might do it better.

Elizabeth Wilner, VP Kantar Media Campaign Media Analysis Group (CMAG), is Contributing Editor at the Cook Political Report