Does the kind of beer you drink or your choice of sport give away your political orientation? If you knew it did, would you hesitate to check in at Friday night’s game?
It seems far-fetched, even Orwellian, but social media data mining was the secret weapon of the 2012 US Presidential election.
From the archives… This piece was first posted on my old website in October, 2012.
Democrat and Republican campaigns are personalizing their messages for you using huge amounts of information, including data you thought was private.
What the campaigns know about you
A New York Times investigation has found both campaigns are tracking the social media activities of their targets and have bought demographic data from companies that collect voters’ shopping histories, gambling tendencies, dating preferences and financial problems.
“Voters who visit religious Web sites might be greeted with religion-friendly messages when they return to mittromney.com or barackobama.com,” Charles Duhigg reported.
By merging social data with public data, marketers have discovered Obama supporters are more likely to listen to smooth jazz whereas Romney backers are more likely to drink Samuel Adams beer and be premium cable TV subscribers.
Romney’s political director Rich Beeson told the Times the focus was on creating persuasive messages.
“(Clothing store) Target anticipates your habits, which direction you automatically turn when you walk through the doors, what you automatically put in your shopping cart,” Mr Beeson was quoted as saying. “We’re doing the same thing with how people vote.”
How does it work?
Campaigns have long targeted groups of voters in broad categories – soccer mums, tradespeople, young families, small-business owners, etc – but recently, campaigns have rediscovered the need to tailor messages.
Media analyst PJ Vogt told NPR that to target individual voters, campaigns had a lot of information about us and smart ways to analyze it.
“Each of us has a specific entry in these databases,” he said.
From public and census data alone, they know your name, address, gender, age, and sometimes ethnicity, and from that data can make assumptions about your socio-economic status and education level. Onto that, they add more data from commercial firms gleaned from surveys, digital tracking tools and social media analysts.
“Now they know that you subscribe to the New Yorker, that you have a weird obsession with sports, that you went on a vacation…” PJ Vogt said.
“They are running algorithms that look for patterns between that big mass of data they have about you with the information their polling gives them about specific attitudes about the election.”
Marketing organisations like Rapleaf take public data – like census and public records – and mash it with surveys and data aggregators.
Already, the public information about you is being compared with your public social network information (your friends list, birth date, profile picture and anything posts you make public). But it doesn’t stop there.
Rapleaf’s Caitlin MacDonald says the problem with social data is that it is “all tell, no show” – it gives marketers information about what people intend to do when really what they want is to track behavior.
She suggests in the future, more companies will adopt the Netflix model, in which everything you watch is fed into the algorithms of the network and viewing suggestions are based on the collective’s choices.
“Companies will need to figure out not just how to extract information about their customers, but how to structure their organizations to act on it in real time … how to segment, price and market effectively, as new information comes in,” Caitlin MacDonald wrote on the Rapleaf blog.
“The way Netflix operates—as a dynamic product that different for everyone who uses it—may foreshadow how leading firms behave down the line.”
Why do they care what you think?
Romney campaign digital director Zac Moffatt says the Republican campaign was moving money from TV advertising to recruiting social media denizens to campaign on their behalf.
“We’ve moved from the ’08 (election) – which was list-building and fundraising – but in 2012, you’re going to use persuasion and mobilisation,” Zac Moffatt said.
“You’re always looking for places to build things that can then be exponentially grown on their own. You want to give people the tools, then they go out (for you).
It’s not just your vote that’s at stake. University of Notre Dame’s David W Nickerson says friends influence each other enormously when it comes to voting.
The demographic theory of like attracting like suggests the friends of those who are open about their political affiliation are likely to vote the same way.
A 2010 German political tweets study found the number of mentions on Twitter for each political party aligned closely with their share of the election vote.
“The mere number of tweets reflects voter preferences and comes close to traditional election polls, while the sentiment of Twitter messages closely corresponds to political programs, candidate profiles, and evidence from the media coverage of the campaign trail,” the Advancement of Artificial Intelligence study found.
Campaign social media report card
Despite these advances, the Pew Research Center’s Project for Excellence in Journalism says both campaigns could be reading constituents better through social media.
The August report found Obama’s campaign had published 614 posts during the two weeks examined, compared with 168 for Romney. The latter managed about one tweet a day whereas @BarackObama and @Obama2012 posted 29.
“Neither campaign made much use of the social aspect of social media. Rarely did either candidate reply to, comment on, or “retweet” something from a citizen-or anyone else outside the campaign,” the Center found.
“On Twitter, 3% of the 404 Obama campaign tweets studied during the June period were retweets of citizen posts. Romney’s campaign produced just a single retweet during these two weeks – repeating something from his son Josh.”