Over the last half-century, consumer marketing science has turned politics into a data-driven business [source]. Ever since the Nixon campaign in 1968, the presidential candidate is regarded as a sort of “consumer product,” where campaign operations market the candidate accordingly for different regions and demographics [source]. Internet technologies gave citizens more freedom to express and associate on massive scales. And they gave the candidate a better understanding of his or her voters. More recently, the merging of Big Data and consumer marketing provided access to even more powerful tools that can micro-target citizens for both donations and votes via meticulously personalized messages. The Obama administration has appointed an official US Chief Data Scientist, DJ Patel, who applied new data and marketing strategies to voter outreach. Similarly, the Hillary Clinton campaign approached a smart targeting company Groundwork, financed by Google chairman Eric Schmidt. Clearly, US politics has become data-driven from campaign to execution.
A Short History of Pioneering Digital Democracy
When it was introduced in the late 1960s, the Internet was mainly used by military and academic communities. For example, PLATO was the first computer network that was designed for computer-based education across colleges. Yet history remembers PLATO primarily for its online community. Activists, presidential candidates, and the usual Washington pundits would use the PLATO communication features to start debates and throw topics on the table. In the early days of the World Wide Web, circa 1991, blogs become more popular. And mailing lists existed since the dawn of email [source].
2004 Elections: From Revolutionaries to Frontrunners
In 2003, running up to the 2004 election, Joe Trippi was the campaign manager for Vermont Governor Howard Dean. Until then, campaign managers were careful about not allowing citizens to speak for the campaign, as they were afraid somebody would say something that would reflect badly on the candidate. However, Trippi spent time enough in Silicon Valley to realize that in the newly-born digital age, citizens were empowered to voice their opinion publicly without need for a mediating middleman. Trippi had to help Dean’s campaign unlearn the old military approach of control and command that, until then, ruled in campaigns.
Trippi understood the power of letting citizens express and make associations aligned to their own agenda, without giving direction or permission. He called it the “largest grassroots movement” in presidential politics. One of his tactics was a blog entitled “Blog For America,” where Dean could interact directly with supporters. Another tactic was DeanTV, an online stream of videos and clips from the campaign.
The then two-year old Meetup.com also played a strong part in enabling supporters to connect and campaign for Dean together. After all, an emergent community formation would be the only possible way to for an unknown insurgent to gain a chance against the presumed frontrunners in DC (in this case, George W. Bush). On top of that, as Internet advisor to the Dean Campaign David Weinberger said “Politics has always been about power, and the campaign is willing to be truly democratic in a way that is really different.”
The number of participants in Dean Meetups ultimately peaked at about 143,000, spread over about 600 locations — a huge number in those days. And engagement in the face-to-face local groups dramatically affected how involved volunteers got with the campaign. The more Meetups people attended, the higher their average donation was to the campaign.
Thanks to Trippi’s novel use of the Internet for small-donor fundraising, “Dean for America” raised more money than any Democratic presidential campaign to that point. Yet all donations averaged less than $100.
Dean ultimately did not win the nomination for Democratic candidate. Yet his pioneering work sparked the advent of a truly data-driven democracy.
2008 Elections: Leaping the Digital Divide
During the Dean Campaign in 2003, Trippi was concerned that elder citizens would not have access to the Internet and therefore could not be targets of his digital campaign. All campaigns had a blog, but they were only mildly popular because the blog applications were not easy to use at that time.
In 2007, running up to the election of 2008, blogging was used massively from the start for the first time. Free expression and association was not reserved to specific blog sites anymore. Every social app you can think of now had features to achieve the same goal. The Obama campaign also leveraged other techniques used by its predecessors such as Trippi’s micro-donation strategy.
But the Obama campaign was groundbreaking in a different way. The advent of Big Data and its use in consumer marketing gave rise to micro-targeting of citizens for both donations and votes via meticulously personalized messages. This could be an email or a banner on your favorite social media app. You wouldn’t find a friend or a neighbor receiving the exact same message. In 2008, the Obama campaign gathered so much information that it was “confident it knew the name of every one of the 69,456,897 Americans” who voted for him, according to journalist Sasha Issenberg.
2012 Elections: Reconciling “Get-Out-the-Vote” Lists with “Fundraising Lists”
The Obama re-election campaign’s 2012 data analytics team was five times larger than in 2008 and hired a real chief data scientist, Rayid Ghani. Soon they discovered there was one huge weakness with their approach: too many databases. Get-out-the-vote lists were never reconciled with fundraising lists. So over the first 18 months, priority one for the campaign was to create a unique massive system that could aggregate all information collected from pollsters, fundraisers, field workers, and consumer databases, as well as social-media and mobile contacts with the main Democratic voter files in the swing states.
The new mega-database didn’t just tell the campaign how to find voters and get their attention. It also enabled the numbers crunchers to run tests to predict which types of people would be persuaded by certain kinds of appeals. Call lists in field offices, for instance, didn’t just list names and numbers. They also ranked names in order of their persuadability, with the campaign’s most important priorities first.
Ghani adopted an approach he developed for supermarkets where they place shoppers in baskets based on their shopping behavior. The most surprising bucket was the one with ‘persuadables,’ which identified shoppers who could be convinced to switch brands for a few cents. He saw a that these shoppers behaved much like swing voters.
“We could [predict] people who were going to give online. We could model people who were going to give through mail. We could model volunteers,” said one of the senior advisers in an interview with Time magazine. The campaign bought ads to air during unconventional programming, like Sons of Anarchy, The Walking Dead, and Don’t Trust the B—- in Apt. 23, skirting the traditional route of buying ads next to local news programming.
Obama’s Chicago campaign office discovered that people who signed up for the campaign’s Quick Donate program, which allowed repeat giving online or via text message without having to re-enter credit-card information, gave about four times as much as other donors. The Obama campaign made the mythical $1 Bn in donations goal, and even went beyond it.
A former Obama staffer, Elan Kriegel, who now leads analytics for the Clinton campaign, suggested the technology accounted for perhaps two percentage points of the campaign’s four percent margin of victory in 2012. In politics, the era of big data had arrived.
2016 Elections: Salesforce.com for Politics and the End of SuperPacs
Hillary Clinton extended the same methodology established by Obama’s team. She also inherited the database they built over two campaigns. Her campaign approached a smart targeting company Groundwork, financed by Google chairman Eric Schmidt and lead by Obama’s campaign CTO. According to Quartz, the goal was to build a “political version of a CRM system (like Salesforce.com) to manage millions of prospect voters (instead of customers).”
“Schmidt’s funding of the Groundwork suggests that 2016’s most valuable resource may not be donors capable of making eight-figure donations to Super PACs, but rather supporters who know how to convince talented engineers to forsake (at least for a while) the riches of Silicon Valley for the rough-and-tumble pressure cooker of a presidential campaign.” [source]
In contrast, the Trump campaign has put a premium not on data, but on human touches made via rallies. “My best investment is my rallies,” Trump said in an interview. “The people go home, they tell their friends they loved it. It’s been good.”
One thing is for certain. We’ll know in the next week whether data again provides a critical advantage.
The campaign professionals in Washington who apply a “command and control” approach and make decisions on hunches and experience is rapidly declining. They are replaced by the work of data scientists and coders who can crack massive data sets.
The combination of “get-out-the-vote” lists with “fundraising lists” into a single actionable database demonstrates that campaigns have become (or call upon) 100% data-driven operations (or services). By doing so, these campaigns gain insights, leaving opponents blind-sided. Data from past presidential elections provides an undeniable correlation: having the best data is vital to turn the campaign to your advantage.
Finally, as these databases are passed on from one campaign to the other, they become increasingly complex. The data scientist teams involved in one campaign may not be part of the next and so the data sources, data lineage, processes, roles and responsibilities, and privacy policies involved may increasingly become a challenge to trace back. Unless they aspire to have a ‘datagate,’ the next presidential candidates may want to think about a data governance platform to provide transparency and sustain compliance through their data science endeavors.
Pieter De Leenheer is cofounder and Chief Science Officer of Collibra. He leads the company’s Research & Education group, including Collibra University, an online learning platform for data governance and data science education. Prior to cofounding the company, De Leenheer was a professor at VU University of Amsterdam. Today he serves as adjunct professor at Columbia University and as visiting scholar at several universities across the globe, including UC San Diego and Stanford.