Why The Candidates Watch What You Buy

It's a beautiful Indian summer day in Paramus, N.J., sunny and 20 degrees higher than locals expect for mid-October. The oaks and maples throughout this bedroom community about 20 miles from New York City are afire with yellow, orange and crimson leaves. It is, in short, a great day to be outdoors.

Which is why Zachary Rynar is having a rough day. Rynar, a 20-something staffer from the Paul Aronsohn for Congress campaign, has reddened his knuckles knocking on doors up and down this residential street. But on a beautiful afternoon like this, nobody is home.

Rynar has no choice but to leave leaflets behind, which more or less defeats the purpose of his showing up in person to talk to voters. Worse, he can't even leave Aronsohn's literature in mailboxes, which would be more convenient. "We have to leave them in the door," he explained. "It's a federal crime if you put them in the mailbox."

Rynar, who's fresh out of Brown, has never gone door-to-door before. His marching orders are to hit the houses listed on a sheaf of papers he was given. Rynar wasn't sure why this list contains some houses while skipping others that are right next door. His guess is that the list is made up of registered Democrats.

This is a tried-and-true method of seeking voters, but it's now considered quaint compared to the cutting-edge, database-driven targeting that both parties are using—though to differing degrees—this year. By now, most all Democratic campaign volunteers like Rynar have heard about the fearsome marketing machine the Republicans are said to be firing up nationally this year. It purportedly slices and dices swing voters into small, identifiable segments—which can then be hit with targeted marketing messages.

It's heady stuff, and Parisa Sabeti, the campaign manager at Aronsohn headquarters a mile or two away, knows it. But Sabeti has the bad luck of being in the Fifth Congressional District (the toupee atop New Jersey, in the words of Stephen Colbert). Even though incumbent Republican Scott Garrett's previously bedrock foundation in the region has eroded somewhat recently, he's still considered safe enough that the Democratic National Committee has given little financial support to Sabeti's team. The Aronsohn campaign raised most of its cash from PACs and unions.

Given Aronsohn's long odds, Sabeti had, for a time, considered going the same route as the Republicans, hiring a firm using a database technology called First Tuesday. But when Sabeti learned that would cost hundreds of thousands of dollars, her campaign took a pass.

Sabeti had a tough decision. On the one hand, Republicans claim to have won Ohio in the 2004 election with targeted database marketing. On the other, there's no independent proof that databases were what did it. But 2004 is ancient history; the real test of the technology will be next week's election. As the Iraq War drags on and headlines have screamed about the bungled Hurricane Katrina effort and the Rep. Mark Foley-page scandal, Democrats find themselves leading by 7 percent in the national polls. In many districts, even in some red states, some Republican officials are clearly in jeopardy of losing their seats in Washington. So the question is: Will cutting-edge marketing technology be enough to save those seats, those Republicans and their party?

THE ELEPHANT'S VAULT

This election will be the first major test for Voter Vault, the GOP's national database. The system's been tried before, but not on a scale this large—or, for that matter, this critical. According to Eddie Mahe, a Washington-based Republican consultant and former deputy chair of the Republican National Committee, the party used a version of Voter Vault in 2002 and 2004. Those databases were used locally, most notably in Ohio in 2004, where some say they enabled George W. Bush to maintain control of the White House by a statistically hair-thin 118,599 votes.

By themselves, voter lists and consumer data are of little use to a party trying to get its candidates into office. It's the synthesis of the two that forms the core of the voter-targeting machinery. A party supplies a list of loyal voters; those voters are profiled for their consumer-goods preferences; then all consumers with that preference profile become targets for the party's campaigning. That group can also be subdivided into endless subgroups, each of which can be targeted with more specific messages—almost always, ones that appear in the mail.

Like many a database, Voter Vault contains standard demographic data such as age, average income and party affiliation. Where VV distinguishes itself, however, is with its lifestyle data. The Vault knows, for instance, a voter's preferred brand of toothpaste and what gym he works out at. Such information allows the GOP to "micro-target" voters by dividing them into categories, each of which can then be targeted with tailored campaign messages.

"What this is about is how you find the individual," says Matthew Dowd, the chief strategist for President Bush's 2004 election, who is now a partner with the corporate brand consultancy ViaNovo, with offices in Austin, Texas, and Alexandria, Va. "You take voter-file information and you combine that with what kind of car they drive, the magazines they subscribe to and their buying habits."

How is the knowledge of whether a given voter drinks beer or wine supposed to help a GOP candidate? Micro-targeting involves predictive analysis. If you buy a lot of peanut butter, a predictive analysis might suggest that you'd be open to a pitch for jelly. The technique has been around since the 1970s when State Farm Insurance, for one, combined its auto and homeowner policy lists and cross-sold customers on both lists.

Political parties caught on quickly to predictive logic. "We knew that Mercury owners tended to vote Republican and Volvo owners voted Democrat," Mahe says. Today, what a customer chooses in the grocery aisle is thought to have a lot to do with whom he chooses in a voting booth.

Lacking a centralized, national database, however, neither party could do much with such intelligence. But now, RNC chairman Ken Mehlman has confirmed in published reports that Republicans have rolled out micro-targeting nationally this year.

Historically, Republicans have been ahead of the curve with database marketing. But the Democrats and former President Bill Clinton's 1996 campaign are credited with pioneering "Life Targeting." For years, both parties had targeted voters based on where they lived and their previous voting pattern. Life Targeting was the first to look at what magazines voters read and what kind of cars they drove, and then tailored political messages toward them.

Alas, in 1996, computers and databases weren't what they are now, and what the GOP has developed trumps their rivals' older model in both scope and sophistication. "If the Clinton plan had been the equivalent of Life Targeting 1.0, President Bush's advisers created Life Targeting 4.0—a quantum leap that allowed them to track millions of voters based on their confidential consumer histories," writes Dowd, with co-authors Ron Fournier and Douglas Sosnik in Applebee's America: How Successful Political, Business and Religious Leaders Connect With the New American Community. "If you're a voter living in one of the 16 states that determined the 2004 election, the Bush team had your name on a spreadsheet with your hobbies and habits, vices and virtues, favorite foods, sports and vacation venues, and many other facts of your life."

PRIVATE EYES

None of this targeting would be possible without the help of private-sector marketing research. In the 2004 election, TargetPoint Consulting of Alexandria expanded the Volvo=Democrat reasoning to considerable proportions. Research found, for example, that Coors beer drinkers tend to lean Republican. Like gin or cognac? Chances are you'll reach for the Democratic lever. In the 2004 election, according to Dowd's book, the Bush campaign forked over $3 million to TargetPoint for political consulting and to conduct surveys (and that was just in Michigan). The same year, the GOP also sent voter lists from Michigan to a data-mining company called Acxiom, based in Little Rock, Ark.

Acxiom has one of the largest collections of consumer data in the country. Acxiom buys its data from a plethora of sources, including airlines, credit card companies, cruise lines and retail stores. As Dowd writes, say a certain customer "subscribed to a wine magazine, gambled at casinos, or collected stamps, Acxiom had a record of it."

When the Republican team submitted its voter list to Acxiom, they were no doubt delighted to receive consumer-related data on 95 percent of the names on that list. Put in other terms, the GOP obtained consumer histories on 5.7 million of the 6 million registered Michigan voters. The list from Acxiom showed "the stage of life (age, marital status, number of children, etc.) and lifestyles (hunter, biker, home renter, SUV owner, level of religious interest) of each voter, drawn from a menu of more than 400 separate categories," Dowd writes.

Next, with the help of surveys and modeling techniques, those Michigan Republican voters were then divided into subgroups with names like "Flag and Family Republicans," "Mellow Bush Supporters," "Religious Independents" and "Terrorism Moderates." These groups become the recipients of highly specialized marketing messages, usually in the form of direct mailings. "You are now targeting based on behavior," says Brian Reich, a senior strategic consultant at Mindshare Interactive Campaigns, a Washington public affairs shop that handles lobbying and ballot initiatives. "You understand a lot more about a person based on how they spend their time and money, rather than on how they identify themselves."

To cite one of micro-targeting's more colorful examples, this year the GOP is going after the 260,000 residents of Michigan who ride snowmobiles (since the state keeps records of licenses, finding out who owns a snowmobile is easy.) Snowmobilers are particularly angry that Democratic Gov. Jennifer Granholm's pro-environmental stance has held up the creation of a trail that would link the communities of Gaylord and Cheboygan, and would likely be receptive to promises by Republicans to take their side on the issue.

But snowmobilers are, if you will, only the tip of the iceberg when it comes to the Republicans' targeting efforts. Saul Anuzis, the GOP chairman for Michigan, says the party this year is targeting 44 segments of the state's population with 12 different messages. He declined to offer more specifics. "Jobs is a big issue," he says. "We'll also be talking about immigration reform."

In preparation for the crucial November elections, this kind of targeting is happening all over the country. "Victory" is the name for the New Jersey GOP's get-out-the-vote drive for both federal and local races. The state's GOP rep Todd Riffle confirmed that the program relies on Voter Vault data, but he refused to disclose details.

What's more, the GOP's targeting reaches beyond seeking out the party faithful; micro-targeting has also identified Demo-crats who might defect if the message is right. "We try to find both political beliefs and traits and other types of behavior, like a Democrat who is behind us on the war on terror and may be persuadable," says Adrian Gray, director of strategy for the RNC. "We want the 'connectors'—people who are leaders of their own social hub and have influence." While Gray won't identify the exact predictive variables, he says they look for Demo- crats with "certain religious interests or outdoor interests."

Dowd argues that finding the influencers is the wave of the future for both political campaigns and for companies. "Instead of using mass media, people will figure out how do I advertise and reach those navigators," he says.

BACK IN BLUE COUNTRY

The Democrats' data-gathering strategy pales in comparison. Prior to this year, some states lost or discarded voter data between elections, according to the DNC. And unlike Republicans, who handpick candidates early on and develop corresponding campaign messages before the upcoming race, Demo-crats often find themselves playing catch-up—waiting until a candidate gets the party nomination, then figuring out how to sell him or her to the voters.

The disparity between these arrangements is obvious. "If you wait until [a nomination] to tell people you have to mobilize, you have a problem," argued Reich. The Democrats are now in a scramble—and the stakes are high. Many political observers are predicting that the GOP may lose its lock on both houses of Congress. But for that to happen, the Democrats must win 15 seats in the House and six in the Senate.

The Dems don't intend to fight those battles with the GOP's type of statistical weapons—at least on as large a scale. The Democrats use two separate databases. One, called Demzilla, is used for fundraising. The party is also building another in-house database for voter targeting.

DNC chairman Howard Dean has indicated that he isn't relying much on database marketing. Instead, the party has taken a decentralized—and decidedly more low-tech—approach, focusing on developing a strong party organization in each of the 50 states.

That decision has frustrated some of the party faithful, who've sought to build their own databases. Harold Ickes, a former Clinton adviser, has set up a for-profit group called Catalist, which is creating a national voter database it will sell to political parties or third-party issue groups known as 527s. The AFL-CIO, Service Employees International Union and Emily's List, a pro-choice women's group, are among Catalist's clients.

The DNC has not abandoned data-driven strategy completely. The party has spent $8 million to rebuild its voter data file, according to Ben Self, the DNC's director of technology. (Self, prior to co-founding the Internet-communications firm Blue State Digital, had been the chief data-analysis expert for the 2004 Dean for America campaign, which was widely cited for effectively using the Web to raise funds and attract support.) The party planned to run modeling efforts in six states this fall, but party officials would not identify the states.

MICRO-TARGETING, MACRO DOUBTS

Given the war and the scandals that have plagued the party in power, the Democrats may believe they have enough momentum—based on public opinion alone—to overcome Voter Vault and its influence. But the party may have another reason for relegating statistical crunching in its strategy. Just like Sabeti, Aronsohn's campaign manager in New Jersey, the Democratic party may simply think that micro-targeting's results don't justify its expense.

Some Democrats wonder what the Voter Vault fuss is all about. "It's ridiculous for Republicans to claim that they've gained an advantage over the Democrats because of their work over the past two election cycles," Self says. "First, the Democratic Party has been micro-targeting for the same amount of time, and second, I'm fairly certain that the Republican Party hasn't developed any new algorithms for modeling that didn't exist before. These are all market research techniques that have been used for years."

Self also argued that the micro-targeting cases that are interesting to talk about—like Porsche owners lean right, while Hyundai drivers tend to not vote at all—are in practice just that: Talk. "To my knowledge, there is no database that has gin drinkers on it, so you have to rely on survey data," he says. "You have to pay someone, and then you have to match your voter file and it gets to very small numbers very fast."

By small numbers, Self means the amount of potential voters that the statistical machinery spits out—not enough, in other words, to justify the effort. Still, small numbers have come to matter a great deal. In 2000, Al Gore missed snaring Florida by a mere 537 votes. (No word on whether they were cognac drinkers.)

Nevertheless, the marketing world thrives on ROI and, so far at least, electoral micro-targeting hasn't revealed much of it. "It's not proven," says Don Green, a political science professor at Yale University. "It's based on speculations about how people with certain consumer traits will respond to different messages."

By contrast, micro-targeting in business has earned its stripes. Walt Disney World, for instance, tapped Acxiom's Personicx database of 120 million U.S. homes to generate profiles of its customers. Using variables including age, income and Internet use, Disney found 14 metropolitan areas that house the majority of its current customers. Better still, the company learned of 10 similar locales it hadn't even known about—a marketing invitation if there ever was one. The grocery chain Food Lion also worked with Acxiom on a segmentation strategy that highlighted 6.8 million target households. Using the data, Food Lion discovered why a program pushing meat in rural stores only worked about half the time. Consumers in the slow-selling stores tended to be older, and hence, not as likely to buy meat as younger shoppers.

Apart from the question of micro-targeting's worth in the electoral realm, some worry that political parties are even using the method. "This can have a chilling effect on participation," ventured Kim Alexander, president and founder of the California Voter Foundation in Davis, Calif. Alexander says that studies show about 25 percent of the population doesn't vote because it wants to keep personal information private.

But the GOP's Mahe countered that there's not much for voters to fear about micro-targeting—not yet. The matched-up data, he pointed out, is mostly used for direct-mail campaigns and hasn't proven valuable for much else. "To try to use it for door-to-door canvassing would be extremely difficult," he says.

Door-to- door canvassing is already difficult enough. Which is a fact that, back on the leafy street in Paramus, N.J., where nobody is home, Democratic campaign staffer Zachary Rynar already knows all too well.