NEW YORK CITY — Campaigns can tailor messages directly to you like never before.
With services such as Facebook and Google providing campaigns with the ability to micro-target smaller levels and with greater efficiency, campaigns are increasingly looking toward a valuable tool to help push voters out to the polls in favour of their candidate.
After the 2016 election, a lot of attention was given to how inflammatory, false, and misleading content on Facebook, Google, and elsewhere online further polarised the electorate. But far less attention was aimed at the use of big data in campaigns, and the effects that the resulting messaging had on political polarization in the US.
It was a topic that Chuck Todd, host of NBC’s “Meet the Press,” attempted to tackle in a mid-March column titled “How Big Data Broke American Politics.”
Todd’s argument was that the misuse of advanced campaign analytics information led campaigns to simply aim for maximum base turnout, rather than aiming for the increasingly small piece of the pie in the center: the persuadable voter. As a result, both Republicans and Democrats have been pulled further to the right and left in their messaging and governing, and they are now responsible only to their supporters, rather than their constituents.
“Why? Big Data — a combination of massive technological power and endlessly detailed voter information — now allows campaigns to pinpoint their most likely supporters,” Todd wrote. “These tools make mobilizing supporters easier, faster and far less expensive than persuading their neighbours. Of course, this isn’t an argument that data itself — be it ‘good’ data or ‘bad’ data — broke the system. It’s how the data was misused and manipulated that brought us to a breaking point.”
The 2016 presidential cycle saw Sen. Ted Cruz of Texas launching his campaign with the firm Cambridge Analytica, a firm that claimed to be able to build a “psychographic” profile of voters. It saw a website designer, Brad Parscale, becoming the most recognisable face of Republican data at the helm of Project Alamo, the Trump campaign data venture that reportedly had “three major voter suppression operations under way” near the end of the election.
After the campaign, Democratic presidential nominee Hillary Clinton suggested that her data operation simply wasn’t as good as the Republicans’, pointing at what she considered a “bankrupt” operation as one of the reasons for her defeat to President Donald Trump.
A number of campaign veterans and data-analytics experts told Business Insider that, yes, there is the potential to misuse data to a negative effect. But each was quick to say that data has not played as great of a role in polarization in comparison to a splintering media, gerrymandered districts, and a tendency for Democratic base voters to pack themselves in major cities, somewhat limiting their electoral power on congressional elections.
To say that the use of big data has led to a more staunchly polarised country “gives some of these people a little bit too much credit for influencing what’s happening in the country and not vice-versa,” Tim Miller, who served as communications director for former Florida Gov. Jeb Bush’s 2016 presidential campaign, told Business Insider, citing the “self-sorting” of voters in urban and more rural areas as well as in social networks that are increasingly less diverse in thought and ideology.
That’s what’s “driven campaigns more towards a kind of rally-the-base turnout model” in recent elections, he said.
“Has that somewhat by, you know, targeted messaging that inflames people’s views on the left and the right? Sure, I think you could argue that it’s been exacerbated somewhat by that,” Miller said. “But it’s also been by social trends.”
“I do think if you look at Facebook, for example, it’s really easy to target people on Facebook with niche messages that kind of inflame their prior beliefs,” he continued. “And then those people share with their friends, who tend to be like-minded, because of the sorting that’s happened in the country. So definitely I think they use these tools to exacerbate it, but I don’t think it’s the cause.”
The tools have changed
In 2008, Democrats were coming off of two consecutive electoral defeats, and the Democratic National Committee was seeking to develop new methods of targeting voters. Publicly available information quickly populated voter profiles. Voter data included points such as vote history, party registration, gender, age, location, and race.
By the time President Barack Obama wrapped up back-to-back electoral victories, with the Democratic data operation given much of the credit for the successes, the GOP had started to invest more heavily in getting its own data operation up to par. The Republicans would spend roughly four years closing the gap between the parties in their respective data infrastructures.
In the final months of the presidential election last year, the Republican National Committee was proclaiming that it had figured out a new way to target on Facebook that, in its early uses, was proving to be successful. And Parscale, the head of Trump’s data team, was leading an operation that was reportedly shaping an increasing amount of Trump’s political and travel strategy, in addition to his fundraising.
Parscale may be called before the House Intelligence Committee soon to answer questions about whether there were any connections between the Trump campaign’s digital operation and Russian officials.
After the 2016 election, Parscale and his team took plenty of credit for both Trump’s improbable win, claiming to see the path to victory form ahead of time. Scott Tranter, a cofounder of the data-analytics firm Optimus, told Business Insider that the biggest misconception about the use of data in campaigns is that “it wins or loses the election.”
“It has a marginal, on-the-edge effect in some close races,” Tranter, who was on the data team for Sen. Marco Rubio’s 2016 presidential bid, said. “So close races matter. The vast majority of races are not close.”
Terry Sullivan, Rubio’s presidential campaign manager, said that although campaigns are able to accumulate data “on steroids” now, the goal of campaign messaging has remained unchanged throughout American history. He cited Abraham Lincoln’s guide to campaigning.
“It’s like a list of 10 things,” Sullivan said.”Basically, identify who is voting, ascertain as to who they will vote for, determine who’s undecided, and persuade the undecideds. This s— hasn’t changed in the following 150 years.”
“Now, the tactics have changed,” he continued. “And the tools. It has become somewhat more scientific.”
‘If you get your base out, you’ll win’
Chuck Todd’s argument — that data has allowed campaigns to aim for maximum base turnout instead of working to persuade undecided voters — was echoed by others, even if they disagreed that it was the use of data itself that was contributing to increased polarization in the country.
Jesse Ferguson, deputy national press secretary for Clinton’s 2016 campaign, said it’s “easier than ever” to identify people who agree with your message and only speak to others who do as well.
“People who want to divide the country are more able to because they can use data to find only the sympathetic audience,” he said. “It is rarely a winning strategy, and it is not a unifying strategy. But it is more available today than it has ever been.”
But he called it “a chicken and an egg” to say whether partisanship allowed partisan data, or if partisan data created partisanship. “There’s no way to say which was first,” he said.
Looking specifically at the past presidential campaign, Reed Galen, who served in President George W. Bush’s administration and worked on Bush’s and Sen. John McCain’s presidential campaigns, told Business Insider that while Trump’s team used data to “stoke” anger, it was clear data was not the difference between a win or a loss.
“Use of data by his guys was making sure to stoke that anger, to hit that adrenaline button to make those people show up,” he said. “And you could argue that Hillary had all the data in the world, and she couldn’t find any buttons to press on people to get them to show up for her.”
Trump ran the antithesis of a data-driven campaign, Galen added, saying that data is “great to have” but “you’ve always got to take the data you get, whatever it is, and layer in some humanity, the personality of the candidate. If you get your base out, you’ll win.”
Scott Tranter, the Optimus cofounder, and Brian Stobie, another cofounder at the company who worked on the Rubio campaign, both insisted that going after a turnout-based message targeting approach is the wrong move. Base turnout is something that can’t be measured until a race is over, they said, while campaigns can track the progress they’re making with undecided voters throughout an election season.
“The analysis is blatantly wrong,” Stobie said. “I think whether it’s any level campaign … 90% of the dollars we think we’re spending on persuasion, not on turnout operation. Turnout is something that we come back to on the Republican side. Democrats are better at doing it in the long term, but Republicans come back to it as an afterthought in the last month.”
“Just think about it in the most cynical terms,” he continued. Data professionals “don’t make money off of turnout. It’s very rare to do that. They make money trying to persuade people.”
Tranter pointed to an increasing availability of partisan media as more a rationale for growing partisanship, rather than the use of data.
“Those guys are running 24 hours a day, and they’re going after a niche audience,” he said. “I’m only going after a niche audience for about 60 to 90 days every two to four years. And I’m using anywhere from $US20 million to $US60 million. Fox News in an afternoon will blow through the equivalent of $US20 to $US60 million. And so, it’s correlated. It might be correlated. I’m sure he found some correlation between voter targeting and polarization. But I’m not sure it’s the causative reason.”
The pair additionally criticised Clinton for pointing at her data operation as a cause for loss. The electoral result, they said, was “a statistical oddity,” not proof of a “bankrupt” data operation, one they said was still a step above what the GOP is able to provide.
“If I had to bet again, I would bet on the Hillary side,” Stobie said. “Three days later or three days in advance. … The Democrats didn’t lose because of their analytics.”
He continued: “What were the factors? No one ever knows, because you just get one result. Reality is that she had a really good data operation. We know the guys who ran it. Reality is we had no good data operation. None of the good Republican data guys were on that Trump campaign. He won in spite of his data deficiencies. Now they’re the victors. Now, she does the stupidest thing on earth and does correlation is causation. They won, ergo they had the best data in existence.”
Tranter added that Clinton likely had the best analytics team, while Trump “had just enough to win.”
Asked about Clinton’s comments, Ferguson said he couldn’t speak to the substance of them because he wasn’t working closely with the data team. But he said that if Democrats “rest on our laurels” while conservative mega-donors such as the Koch brothers and the Mercer family continue to invest heavily in Republican data operations, Democrats will “wake up in January of 2019 with Paul Ryan still having the gavel and January 2021 seeing Donald Trump heading back into the White House.”
Andrew Therriault, formerly a leader on the DNC’s data operation, publicly criticised Clinton for the comments in their immediate aftermath. Now working for the city of Boston, Therriault pointed to the Todd’s argument as a “stretch” lacking any “thesis.”
“The claim seems to be that, basically, because data allows campaigns to target their supporters, that therefore there isn’t an effort to persuade, and that has been the cause of polarization,” he told Business Insider. “And I think the supposition in there — that I think is just a stretch, to put it mildly — is that campaign persuasion, until recently, was what kept us from being polarised. And that seems like a huge assumption that there is no thesis for at all.”
Therriault wrote a Medium post in February titled “We Shouldn’t Blame Data for Bad Campaign Messaging.” “Data can’t salvage bad campaigns, but it can help good ones,” he wrote. “And for us to recover after 2016, we’re going to need all the help we can get.”
He said a lot of focus on this issue has come about because of Clinton’s failed campaign.
“That’s where all of this comes from,” he said. “And there are questions to raise regarding their messaging strategy. But that’s not about data, that’s about messaging. Data is not a strategy. And data is not a replacement for strategy.”
For Therriault, figuring out who is still persuadable, in a climate where he said roughly 10% of voters are “truly independent,” is still “the holy grail of data.” Still, the past election was one in which a lot of what was assumed about the American electorate was thrown out the door.
“Obviously, this past election was anything goes,” he said. “There were a lot of rules that just don’t apply anymore.”
Pointing to gerrymandering and the creation of extremely safe partisan districts as more a cause for partisanship than data itself, Miller compared blaming the latter to blaming a gun manufacturer after a shooting.
“People are much more concerned about primary challenges for the first time, and pleasing their base, than they ever were, rather than a general election,” he said. “Sure, you can” misuse data. “It’s also like blaming the gun manufacturer. Sure people use voter data to target things, but that’s not the reason it’s happening.”
Taking aim at data assumes the premise that paid media, and not earned media, is driving partisanship, Miller said.
“And it’s not, it’s the exact opposite,” he said. “It’s earned media. There’s so much more earned content than ever before. And let’s be honest, you don’t really use data to target the earned media the same way.”
Sullivan agreed, citing Trump’s election as proof.
The Rubio campaign measured earned and paid impressions from the day Rubio joined the campaign until he dropped out of the race in March. Sullivan said there were just two weeks from the day Trump joined the race that he did not have more earned impressions than every other candidate combined.
“And so the sheer volume in his ability to drive earned media just shut out everything else,” he said. “What you see out of that is that wasn’t about using data to target anything. We were able to measure it, but the fundamental was that Donald Trump didn’t use data at all. He just knew how to get the media to take the bait over and over again. He was PT Barnum. That is the exact opposite of using analytics to target your message.”
Business Insider Emails & Alerts
Site highlights each day to your inbox.