It was the winter of 1982, and I needed a job. Two years earlier, I had worked on the unsuccessful reelection campaign of Sen. Birch Bayh (D-Ind.). Prior to my work on Bayh’s campaigns, I had spent many years in the U.S. Senate compiling lists the Senate computer services would use for Bayh’s franked mail program.

So when I was approached early in 1982 by a company tasked with acquiring voter registration files that could be used to match against the AFL-CIO membership and identify whether or not members were registered to vote, I was intrigued. Once compiled, those lists could be sold to political campaigns.

The job piqued my interest for a few reasons. First, I knew from experience that finding computerized databases hadn’t exactly been easy in the 1970s. So I was intrigued about the prospect of doing this in the political world. During Bayh’s reelect in 1980 we certainly didn’t have the benefit of computerized voter lists. Any campaign with a large canvass, phone or mail effort would be greatly assisted by having access to lists like this. Not only would the campaign be sure it was contacting registered voters, but it might have the ability for enhanced targeting—by party, age, race or the likelihood to cast a ballot.

So I took the job and began what has been a more than 30-year odyssey of working with voter files and being an eyewitness and participant to a part of campaigning that has virtually exploded in American politics. In many respects, those of us working in this field at the time were inventing a new business—one that would grow and change as modern technology assumed the dominant role it now occupies in our daily lives.

The computerization of voter files

As this new world of political voter files developed, the computerization of America was still in its infancy. As it matured, so did the voter file business. After looking back on 32 years in this game, I can try to articulate the vast changes we’ve seen over that period of time, and to understand exactly how the face of campaigning has changed as a result. These three decades have been marked by tectonic shifts in the way politics and political communication is conducted. More specifically, three major developments have brought us to where we are today.

While creating an inventory of computerized voter files during those first years in the early 1980s, it became clear that there was beginning to be progress toward greater computerization of these government run databases. As that inventory of information was updated each year, increasing numbers of town, county and state voter lists were computerized. Government officials were quickly learning about the economic reasons to computerize their databases, as well as the obvious efficiencies and time saving. As these databases were being made available to political candidates, those who were elected took office with a greater understanding of the benefits.

The problem: It was a cumbersome process. It required expensive mainframe computers—machines that could only be operated by trained programmers and required large investments to buy or rent the hardware and software, as well as the physical plants necessary to house them. When I entered the business, it was into a company that had its own mainframe. I would never have been able to start this business on my own.

In 1984, while I was just getting my feet wet with computerized voter files and the selling of voter lists, labels and direct mail to campaigns willing to use them, I found myself pitching a campaign that would change the entire way I approached the business. Then- Rep. Al Gore was running for an open U.S. Senate seat in Tennessee, and he wanted computerized data for the campaign. Gore’s campaign manager, Hal Malchow, hired my company to build a statewide voter file.

From a combination of typed and mimeographed precinct lists we had to keypunch to a myriad of computer media—from floppy disks in many sizes to 9-track tapes and 8-track cartridges. The database was built and telephone numbers appended by an outside vendor. I suggested to Hal that the Tennessee Democratic Party ought to be convinced to make the database theirs. By having the state party make the data available to candidates at all levels, it would save the Gore campaign money, put the state party in the business of providing services to its candidates, make computer data available to candidates for the first time and allow a vendor like me to work with candidates I would have never been able to reach otherwise. To me, it made sense for all concerned.

That same year I convinced the Indiana Democratic Party to take on the same kind of project, and in the next few cycles I was able to convince a growing number of Democratic state parties to do the same. These projects became the central part of our business, which grew even larger when the Democratic National Committee began encouraging state parties to have their own voter file managed by a reputable vendor.

So in 1991 I put up my own shingle, opening Blaemire Communications and soon began to sign up more Democratic state parties as clients. We also worked more and more with political consultants across the country who were quickly beginning to understand how high-quality databases could enhance voter contact efforts for clients. The more professional the business became, the more its processes improved and the common mistakes we had been making decreased.

As the decade moved on, our work got better, more campaigns understood they needed high quality voter data, and more consultants depended on firms like mine to help them with their political clients. So many campaigns on our Democratic side now had access to voter data because of the large number of Democratic state parties with voter file projects making that data available to candidates at all levels, and thanks to the proliferation of political consulting firms. By the second half of the decade, however, the changes that had seemed incremental soon came upon us with a fury.

The rise of the PC

Anyone working in a computer related business in the 1970s relied on mainframe computers for processing of any kind. That all changed with the introduction of the personal computer. I received my first IBM desktop computer in 1983, which was my first full year in the voter file business. With copies of WordStar and Lotus the only software I had to learn, along with the Microsoft DOS operating system, it was my early introduction to using a computer without depending on a programmer.

For most of the decade, major database processing was still done on a mainframe and those mainframes were operated only by trained programmers. Having a business providing political computer services meant working in mainframe facilities until the mid-90s. By then, there were simply too many people working in all manner of businesses using PCs, so expecting everyone to depend on large and expensive mainframe computing was becoming unrealistic. PC networks soon became available and affordable, allowing people like me to transfer the business out of the mainframe environment to one that was less expensive, more nimble and efficient.

It was 1997 when I moved my company out of the mainframe facility where I was an in-house client and tenant to space of my own with my own hardware and software. The financial requirements were now tens of thousands of dollars instead of hundreds of thousands. My company’s practice had been to take our internal costs charged by the mainframe facility and mark them up, hoping the markups were adequate for a successful business. Now, even though my overhead went up, my per-job costs virtually disappeared. Pricing could be more flexible. The higher rent and payroll became less important than our new ability to be creative and truly entrepreneurial. The PC network had unleashed our company, as it would for so many other companies in this business and others.

Into the Internet age

My company thrived during the first years away from the mainframe. We had become the principal company managing state Democratic Party voter file projects and were providing data processing services to most of the large membership organizations in the progressive community. A large portion of our costs were for the computer media our data would be copied onto, magnetic tapes and diskettes, and the materials required for printing Cheshire labels, pressure-sensitive labels, laser-generated walk and phone lists and laser-generated direct mail. By 2000, 90 percent of our revenue was printed matter. Data transmitted electronically was 10 percent of our business. Two years later, those numbers were reversed.

We were now able to email phone files to phone bank firms or to the campaigns themselves. The transmittal of data was faster than we could have ever imagined. Our material and shipping costs virtually disappeared while our turnaround time to fill orders for clients sped up exponentially. Clients were happier and profits were higher.

In 2002, we also learned that a few state Democratic parties were using online systems that allowed campaigns to access voter data directly. As much as I was comfortable with the way we were serving the Democratic and progressive community with voter data, I was soon convinced that we would also have to create an online system. Access to voter data had always come through professionals like us. We could make sure mistakes were not made and those understanding the data best were still in the best place to serve clients. But the toothpaste was out of the tube, so to speak. People were clamoring for the ability to access the state party voter files themselves. It was no different than the way we were now buying books through Amazon. The middle man’s existence was in danger and without our own online system we would be standing alone as a middle man.

During 2003 and into 2004 we created our own system called Leverage. That fall, we hosted voter files online for several of our Democratic state party clients, as well as for the DNC. Our online presence in the Democratic community was second only to Voter Activation Network (VAN), one of the firms that has been doing this at least a cycle longer than us. Access to voter data was now becoming truly democratized. Just as the PC network had allowed us to operate much of our computer services business without programmers, now the only requirement necessary for hands-on access to our databases was minimal knowledge of a PC.

Instantly, the number of people with hands-on access to voter file databases went from a few dozen to thousands. This changed everything. New campaign staffers would consider their state’s voter file as the VAN, or as Leverage, or one of the other systems competing in the space. As much as I would argue that VAN and Leverage were interfaces for accessing the database, and that the databases were best created by Blaemire Communications, few wanted to listen to that. The expertise required to build and enhance a voter file properly was something taken for granted. What campaigners saw was what the interface looked like—they made assumptions about the data inside.

After the 2006 election, the DNC decided to build voter files for the state parties itself, hosting all of them on VAN. Because they were free, it spelled doom for my company, which had contracts with 26 Democratic state parties to build and manage their voter files. If that wasn’t bad enough, many of the progressive organizations that represented the remainder of our client base got together to create a company to build a national voter file database they could depend on. That company became Catalist, and my non-party clients were soon gravitating its way. Within a year, I would gravitate that way myself, merging my company with Catalist in December 2007.

The advance of predictive modeling

We have always tried to use information we know about people to predict their behavior, both in marketing and in politics. But now we could take the massive amount of information accumulated on voters and model their political behavior as a result. It’s the third major development that has reshaped this business over the three-plus decades I’ve been in it.

Techniques to model behavior are now mind-boggling. Our data constitutes a devastating mix of ingredients for creating a rich soup of information, always growing and being refined. We are overlaying a growing list of predictive values to the large amount of individual data already made available on voters. Not only are we able to choose to contact a voter because he or she is a Democrat or a Republican, or old or young, or a good voter or not, but we can also choose from a number of models. We are scoring voters on their likelihood to take an action in support of a progressive issue, or to donate. A voter is similarly scored on his or her media consumption habits and whether it is better to contact that voter by mail, phone or online. All of these are just examples of the many things that have been and are being modeled because of the advances in modeling technology and the way the modern campaign accumulates data. No longer is the data thrown away.

No longer is this just the profession of trained programmers. The growing amount of artificial intelligence and known individual data is a powerful combination that makes campaigns smarter. The world of political computer services looks nothing like it did when I began. It’s hard to imagine what another 30 years of progress will bring.

Bob Blaemire is the Director of Business Development at Catalist.