GOP embraces big data

Teradata is just one of the major data outfits with which leading Republican strategists are talking in their declared effort to match Barack Obama's big data campaign tactics, according to one person with knowledge of the strategy discussions.

The Republican National Committee would neither confirm nor deny talking with Teradata, but was emphatic that no deal is in place. Teradata also declined to comment. There's unlikely to be any final deals until the RNC appoints a chief technology officer, which it has pledged to do by May 1.

But if Republican strategists are still shopping for formal partners, their goal is clear: a new, more open database that will make it easier for Republican candidates to share what they're learning about voters — and for the party to share voter information with technology developers in order to build apps for use in coming campaigns.

"At lots of levels, for lots of reasons, there's a lot of people that we're talking to," Mike Shields, the RNC's new chief of staff, told ProPublica.

Both Republicans and Democrats already have databases of basic information about every voter in the United States. But Obama's campaign made big strides in connecting data from different sources, like campaign donation records, consumer data and volunteer lists, in order to produce more detailed profiles of individual voters.

The Democratic National Committee has also streamlined the way information flows between local volunteers and the national party, so that data about voters collected by many different campaigns — such as a Minnesota voter's stance on gay marriage or whom a Virginia voter supported in a state senate race — allends up in the same database in D.C.

Republicans want to match these innovations — especially the flashy ones, like the Obama campaign's ability to link people's Facebook profiles to their official voting records. They also want to use data to make predictions about individual voters, not only about how to influence their vote, but about how to maximize their potential political donations.

This is where Teradata could certainly be useful. The company is not a data broker, an outfit that strictly sells information about consumers. (So, for instance, the GOP wouldn't be getting any of Apple or Wal-Mart's data.) Instead, Teradata helps companies organize their own data, so that they can pick out unexpected trends — for instance, that Wal-Mart shoppers stocking up for a hurricane often buy strawberry Pop-Tarts.

When working with Isle of Capri Casinos, Teradata built a system to combine information about customer gambling habits with data from the company's hotels. The new system sends an automatic alert to casino hosts whenever a "high-value guest" arrives at a hotel. It also tracks how different customers respond to coupons, emails, and special offers.

This kind of detailed tracking has become ever more central to data-driven political campaigns. Almost every day, Obama's re-election campaign tested 12 to 18 different email variations, before sending out the best-performing fundraising email to its entire list — a testing strategy that sometimes earned the campaign an extra million dollars, or more.

The campaign also tracked individual responses to email blasts — storing information on whether someone had, for instance, signed a card wishing Michelle Obama a Happy Mother's Day, and using that information when asking the same people to sign a birthday card for Barack.

Obama's data team also generated individualized predictions about voters. The team calculated, among other things, which people were most likely to be persuaded to support Obama based on a conversation about a certain policy issue — information that then allowed field organizers to be more strategic about the houses they visited and the phone calls they made.

Working with a company like Teradata would only be a first step toward this kind of sophisticated data program. Obama's 2012 campaign considered using Teradata, but ended up going with Vertica, a Teradata competitor, paired with open-source software Hadoop, to organize and search through their huge quantities of data. But, as former Obama staffers point out, having masses of information doesn't do anything on its own: You have to use the data to ask the right questions.

Wal-Mart famously used its database to ask what products customers tended to buy before hurricanes. The Obama campaign used its data to ask whether the voters it wanted to reach were watching the evening news — or other kinds of television shows altogether. The campaign used the television-watching data it acquired to figure outexactly what shows the voters they wanted to reach were watching, all of which made for more cost-effective ad placements.

The result? The Obama campaign bought more targeted ads, while spending less per television spot than the Romney campaign, according to data collected by Kantar Media's Campaign Media Analysis Group.

The campaign also constantly adjusted its predictions — and checked on the big picture of the campaign — by connecting voter information with detailed polling data.

Shields, the RNC's new chief of staff, called the data developments "a space race" between the RNC and the Democratic National Committee.

"They put up Sputnik, but there's no reason that we can't put a man on the moon, and leave them behind," he said.

As well as hiring in-house data analysts, the RNC plans to make it easy for outside software developers to access the party's national database. The goal, Shields said, is to create a "vibrant marketplace" of digital tools and applications that developers can sell to Republican candidates — all based on the party's own voter data. Think about the apps that connect to your Facebook profile — but for politics.

If the plan takes off, some of the GOP's closely-guarded voter data will soon be available in new ways. Obama's 2012 canvassing app, which anyone could download, included a map of the user's current location that displayed the first names, addresses, ages and genders of nearby Democrats.

The RNC will still get to control which developers are allowed to access its data. But its plans for a more open data platform will require that the Republican establishment confront technical, legal and cultural hurdles.

"[The RNC] is an organization that is trying to figure out where they sit with technology in general. They're going to have to make an investment in a big way, if they're going to go on with open development," Harper Reed, the Obama campaign's chief technology officer, told ProPublica.

The hard part about opening up your data is trusting the users, Reed said — including the users you don't like. What happens if some Republican developers want to use Republican data to build a pro-choice app?

"This would be a challenge for any organization, not just a political one," he said. "It sounds interesting. It sounds innovative. It's a challenge."