Election 2012: Who's Winning Big Data Race?

With one week left until the U.S. presidential election, it may be too early to say conclusively whether the Obama or the Romney campaign has more effectively used big data analytics to help land the most powerful job in the world.

But a handful of organizations have been trying to do just that. And despite coming from nearly opposite ends of the philosophical divide between public interests and those of private industry, they are coming up with similar conclusions.

Judged purely on the basis of how well the campaign uses data analysis to communicate its messages most effectively -- not the content of the messages themselves -- President Obama appears to be the clear winner.

[ For more on how the campaigns are using big data, see Big Data Chases Election 2012 Undecided Voters. ]

The Obama campaign has been far more effective at identifying potential voters for email promotional campaigns, and thanks to the its wide base of recipients, it has been more successful at getting those voters to accept the emails. It's also been successful at soliciting donations through email campaigns directed via sophisticated analytics, according to email marketing analytics provider ReturnPath.

On the other hand, email from the Obama campaign is also far more likely to be considered spam (5% vs. 0.8% for Romney), more likely to be thrown away (11% vs. 8%), and much less likely (0.04% vs. 6%) to be forwarded to another voter or email address, according to a ReturnPath analysis of activity in two million email inboxes between Aug. 27 and Oct. 10. (Because so many voters are willing to accept email from the Obama campaign, even with a higher percentage being rejected as spam, the overall number of accepted emails exceeds that of the Romney campaign.)

With an email list one-fifth the size of Obama's and almost half of all recipients routing Romney emails directly into the spam folder, the Republican candidate is at a distinct disadvantage, at least when it comes to campaigning through email, according to ReturnPath.

The differential between Obama and Romney's email-marketing stats may be misleading, however.

The Romney campaign hired Ft. Worth, Texas-based consumer analytics firm Buxton Co. to identify targets other than the largest number of potential voters and donors, according to an Associated Press story in August. Instead, the Romney campaign focused on finding voters and donors who had been previously overlooked by both Democratic and Republican campaigns. Specifically, according to the AP, the Romney campaign had Buxton sift through data on voters in heavily Democratic districts, looking for those with conservative sympathies, high incomes, and presumably, a sense of having been forgotten due to the comparatively left-leaning population around them.

"An early test analyzed details of more than 2 million households near San Francisco and elsewhere on the West Coast and identified thousands of people who would be comfortably able and inclined to give Romney at least $2,500 or more," according to the AP story.

In the heavily Democratic San Francisco area alone, the Romney campaign brought in more than $350,000 in donations at an average of $400 per donation. The Obama campaign, in announcing an unusually high $181 million in donations during September, pegged the average individual donation at $53 and said 98% of all contributions were for $250 or less, according to Politico.

Obama campaign donations topped have $1 billion since January 2011, according to Politico, compared to $921 million for Romney. The figures include money raised by both political parties as well as individual campaigns.

Big Data Beyond the Inbox

Of course, successful fund raising and email marketing campaigns do not by themselves a president make.

The Obama campaign appears to have relied on big data and analytics to even a greater degree than the Romney campaign. In 2011 the Obama campaign advertised for direct-response marketing professionals, analysts and data geniuses, and other titles more common to customer research and database marketing than politics.

The result has been success at building support via email campaigns and social networks. The campaign has also used strategies such as profiling millions of Internet users into targeted categories for mass emails or ad campaigns, according to ProPublica.

If the Obama campaign's successful use of social networks and Internet marketing was the leading digital innovation of the 2008 campaign, its giant database of voters (code-named "Project Narwhal") is the campaign's leading digital innovation for 2012, according to a February article in Slate.

For the 2012 Obama campaign, more than 100 data specialists, predictive modelers and other "data geniuses" cleaned and verified the contents of the earlier database, combined it with voter registration files, information from Facebook, the Democratic National Committee and other sources into a comprehensive set of data on 23 million voters and 25 million additional Obama Facebook fans, according to the Guardian.

The resulting database contains such comprehensive information on voters and volunteers it prompted even Obama campaign staffers to raise privacy issues. Most controversial were practices such as collecting data on Facebook friends of Obama campaign volunteers and donors. Ultimately, however, the campaign defended the mega-data set as the most efficient way to find and connect with voters and to solicit volunteers. "This is the Moneyball moment for politics," said Obama staffer and campaign blogger Sam Graham-Felsen in the Guardian.

The Romney campaign, while more secretive about its promotional and analytical efforts, is aggressive about tagging, following and serving Romney-friendly ads to Internet users, using persistent tags, cookies and other behavioral-marketing techniques, according to ProPublica.

The Romney campaign is also aggressively targeting voters via giant databases, not only from Buxton Co. CampaignGrid, a data and analytics provider that works mostly for Republicans, is helping the Republican candidate by using its database on 110 million voters -- 65% of all registered U.S. voters, the Guardian reported.

This close to the election, however, the focus of politically minded data analysts is shifting to the outcome rather than the process.

The data-driven reporting group within the Associated Press, for example, have already begun posting predictive summaries it calls "snapshots" of the current status of the candidates. It creates these using polls and official voting data to project the number of electoral votes likely to go to one candidate or the other.

Despite the enormous volumes of data that's been crunched, mined and analyzed throughout the runup to the election, the only meaningful numbers are the latest ones from the AP. Unfortunately for partisan analysts and big data fans, there's no way to know for sure how accurate all the information and predictions collected by each campaign actually is.

For that -- along with a final analysis of the effectiveness of big data-driven campaigns -- we'll have to wait until a week from Tuesday.

In-memory analytics offers subsecond response times and hundreds of thousands of transactions per second. Now falling costs put it in reach of more enterprises. Also in the Analytics Speed Demon special issue of InformationWeek: Louisiana State University hopes to align business and IT more closely through a master's program focused on analytics. (Free registration required.)