Victory Lab

For Sale: Detailed Voter Profiles

The RNC and DNC are getting into the data mining business. Will they trade sensitive, strategically valuable information for an influx of cash?

Karl Rove
Karl Rove


The big story of this campaign season has been the rise of Super-PACs, but they are not the only quasi-independent power that could redefine the modern political enterprise. Both the Democratic and Republican national committees have embarked on plans to develop data hubs in the hopes of becoming players in the vibrant private-sector marketplace for voter data. Party bosses have collected information about voters that interest groups like labor unions and the Koch-funded FreedomWorks would pay big money to access. Will the parties be willing to trade their most sensitive, tactically valuable data for an influx of cash?

For decades, the most prized asset a state Democratic Party owned was its voter file. In its simplest form, a voter file is a roster of registered voters assembled from the rolls of local election authorities. But state parties were able to add reams of individual-level information gleaned from years of interacting with voters: their phone numbers, volunteer histories, and pet issues. In some states, the voter file provided enough texture to offer an ethnographic lens on local activist culture. The New Hampshire file, for instance, flagged individuals who had displayed lawn signs or brought food to a campaign headquarters to feed volunteers. This was information that campaigns could not get elsewhere, and party bosses put a price on it, either selling their voter file to candidates or saving it as a prize that could be extended only to those they endorsed—often a crucial way of protecting incumbents or playing favorites in primaries.

After 2000, however, Democratic strategists at the national level came to believe that having their party’s voter data divided across 50 different fiefdoms—often maintained in distinctive formats, accessible only by incompatible software systems—limited its value. The world of campaigns was undergoing a major shift from looking at voters primarily at the precinct level to profiling them according to individual attributes. Computing power had improved dramatically, allowing campaigns to process voter files in new and productive ways, especially when they could be mashed up with records from commercial data vendors that documented individual buying patterns, memberships, and subscriptions. As the 2004 election season approached, DNC chairman Terry McAuliffe offered state parties a deal: If they shared their files with the DNC, the national party would return the records brimming with new personal details acquired in bulk from commercial data vendor InfoUSA.

McAuliffe’s database was largely a flop, and only under his successor Howard Dean did the party realize its goal of developing a workable national voter file that presidential candidates and state parties could use. But it quickly became evident that a national party was a poor custodian of such an asset. A list of nearly 170 million adults—larger than nearly all commercial databases except for those maintained by credit-rating agencies—required computing power and engineering expertise that could never be satisfied by party budgets that had been reined by new campaign-finance laws. Even once a list had been developed, those same regulations limited a party’s ability to share its resource with its allies.

On the Democratic side, some of those allies set out to develop their own voter file, outside the Federal Election Commission’s oversight. In 2006, former Clinton aides Laura Quinn and Harold Ickes—among a cadre of old party hands suspicious of Dean and his “50-state strategy”—raised $5 million from private investors, including George Soros, to build a private data warehouse with records for the entire voting-age population. Called Catalist, the theoretically for-profit company imagined itself as a public utility, with less interest in returning a profit to its investors than becoming an indispensable tactical resource for the American left.

Their customers included the major labor unions, women’s and environmental groups, and occasionally campaigns who considered the DNC’s database insufficient—including Barack Obama’s in 2008. Catalist did the basic work of stitching together lists from local election officials, but some of the most valuable data came from its customers. The company described itself as a consortium, and every contract required a customer to contribute something of value back to Catalist. Rock the Vote used Catalist to identify adults it could target as part of its registration drives—in exchange, it put personal information gleaned during those drives back onto Catalist’s servers. Other Catalist clients would add data points from their own interactions with those voters: EMILY’s List flagged some as pro-choice, the Sierra Club marked others as donors. Those touches of individual detail helped to form well-rounded individual portraits of political behavior that didn’t exist on file at the board of elections. After a while, algorithms could mine those portraits for patterns for so-called look-alike models: What traits predicted whether someone was a likely donor to an environmental cause? Or an unregistered pro-choice voter?

Political data was being converted from a commodity bought in bulk to a boutique creative product, and Catalist created such a robust demand for it that the DNC has embarked on an ambitious plan to win back some of their lucrative business. In an email to liberal interest groups and consultants earlier this month, party officials announced that they had partnered with one of Catalist’s competitors, TargetSmart Communications, to make the party voter files available to lefty and nonpartisan groups on a state-by-state basis. (All but a handful of state parties have signed on thus far.) This new Voting List Management Cooperative—which Democratic operatives are calling “the Co-op”—offers electoral data that’s being marketed as fresher than Catalist’s. The freshness claim is based on the fact that political parties still have a privileged place in reading the electorate: Party activists take stock of who has moved or died before local election authorities get around to pruning their files and they keep the most current updates on who has voted early or by absentee ballot. (Some local authorities release that information to parties before it’s available to the broader public.) A share of each purchase from the Co-op will be sent back to the state parties, which helped to create and maintain the lists but never before had a steady channel to sell them to a broad audience.

“Making them accessible on a national level creates a whole new revenue stream,” says New Hampshire Democratic boss Ray Buckley, who leads the Association of State Democratic Chairs. “It’s not costing us anything. It has the potential of a good return or an awesome return.”

But despite the prospects for new revenue, Democratic leaders have been wary of relinquishing control over some of their most valuable resources to paying customers who could use it to undermine the party’s interests, particularly in primaries. The DNC will keep some of the boutique data it manufactures— statistical models for candidates and parties to use—out of the databases that most Co-op clients will be able to access. And the Co-op will closely monitor its clients, too: Under its rules, state party chairs will have a veto over each sale, allowing them to deny rival operatives’ or dissident unions’ use of their data. Indeed, party support is at the heart of the sales pitch the Co-op is making to customers currently paying Catalist subscriber fees: If they switch, their money will no longer go into a private company’s coffers but help to fund party activities. It is an appeal that may be more persuasive to consultants and pollsters who rely on state party chairs for their business than outside groups, which prize their autonomy and generally feel they owe little fealty to party bosses.

“We are not Democrats and not all of our members are Democrats,” says Mike Podhorzer, the political director of the AFL-CIO, which was one of Catalist’s first customers and relied on the company’s data when it backed a liberal challenger to dislodge incumbent Democratic Sen. Blanche Lambert Lincoln in a 2010 Arkansas primary. “A lot of other groups are in the same position. Years ago, when Catalist started, that was part of the reason for doing it. There was a sense, at least for progressives, that having our own independent source of data was important.”

Now Republicans, who developed their own national voter file 15 years before the Democrats, are scrambling to come up with an apparatus to share and market voter information. Like the Co-op, the Data Trust recently established by the Republican National Committee will be a hybrid, a private company that party bosses built but can’t formally run. “We’re very much in the same boat as the Democrats, looking for a better alternative in terms of how to manage our database,” says former Michigan Republican Party boss Saul Anuzis, who now chairs the RNC’s technology committee.

In the aftermath of 2004, Karl Rove looked warily upon Catalist’s efforts to build an apparatus outside the national party leadership to house the data for targeting voters. “I thought at first it was a comment on their inability to raise money on the national level, that they had to do it on the outside, but it may have been that they were just ahead of the power curve in figuring out how to do it,” Rove says of the Democrats. “I think in 2009 and ’10 they did move ahead of us, because of the inability of the national party to keep pace.”

Rove, now the chief strategist behind the American Crossroads Super-PAC, finds himself in much the same situation Democrats did six years ago. His party was outmaneuvered in the last presidential election, and responded by anointing a party chairman so distrusted by party elites that many reconsidered their thinking about the value of outside institutions to manage an entire movement’s tactics. (The role of Howard Dean was played by Michael Steele.) When Rove first convened a group of Republican operatives to his home on Washington’s Weaver Terrace nearly two years ago with the goal of designing a single electioneering machine to consolidate the right’s fragmented interest groups, Catalist was an obvious model.

In designing the Data Trust, Republicans have struggled to build a party-chartered independent business that allies will want to use. The terrain is perhaps even more precarious than it was when Catalist was formed; Tea Party challenges to Republican incumbents have shown how easy it is for the party’s interests to diverge from those of its interest-group. Negotiations within the RNC had focused on how much influence should go to outside players —like Rove’s Crossroads or the network of groups funded by the Koch brothers—and for now the party’s interests have won out, settling on a model closer to the Democrats’ new data Co-op than to Catalist.

The RNC has structured the Data Trust as a private company with as much party DNA in it as possible: Its CEO is former RNC chief of staff Anne Hathaway, and three seats on its have been given to former committee chairmen: Ed Gillespie, Jim Nicholson, and Mike Duncan. Party officials are insistent that the arrangement is only a short-term contract that can be easily severed: Because the deal is for a list share, the RNC can pull its data out on a whim and quit the trust for good.

While the new company will add commercial data to party files, there is little ambition for it to develop the data-mining expertise to work through it. In fact, the entire Data Trust staff is likely to be smaller than the number of Catalist statisticians who work only on developing what they describe as data “synthetics.” “Our philosophical model is a little more open-source,” says Anuzis, the RNC technology committee chair. “It doesn’t make sense for the Republican party to duplicate what the market can do.” Given the importance, and value, of data in this election, the Republican faith in markets will be tested in more ways than one in 2012.