From an international perspective, the rise and rise of Donald Trump has dominated the narrative of the 2015 US presidential election. His ongoing success at attracting supporters has been called an “anomaly” for many reasons. One is because his primary campaign is, if not entirely then to an unusual extent, self-funded. Another is that while other candidates have all developed analytical strategies for targeting swing voters on the campaign trail, Donald has not.
In fact, he has been vocally critical of analytics and Big Data in election campaigns. Obama’s data-driven strategy has been credited by many as integral in his 2012 election victory, and set the blueprint for the majority of today’s political campaigns. Trump was dismissive, however, telling AP “I’ve always felt it was overrated. Obama got the votes much more so than his data processing machine. And I think the same is true with me.”
Trump’s strategy has instead been to use social media, and the conventional media coverage which follows it, to gain exposure through controversial opinions. He has publicly referred to his social media accounts as being worth “$2 billion in free advertising” to him.
Evidently this has served him well during the primaries. But it seems likely to me that when it comes to fighting a presidential election, a different story will unfold. While he will have the option of leveraging data collected by the central Republican Party which he represents, the lack of his own data strategy may leave him at a distinct advantage to his more analytically-aware opposition.
In an analytically-driven election campaign the focus is on targeting swing or undecided voters. Why waste time campaigning to those who are definitely going to vote for you, or those who never will in million years? This technique was pioneered by the Obama campaign in 2012 when a team of over 100 data analysts were tasked with running over 66,000 computer simulations every day.
First, Obama’s analysts in The Cave collected and amalgamated all the data they could from voter registration data, donations, public records and bought-in third party commercial data (including data mined from social media).
The team then identified three variables which they believed could influence the election, and which could in turn be influenced by their campaign. These were voter registration numbers, voter turnout and voters’ choices.
Armed with their sophisticated demographic information, targeted campaigns were then launched. These had the aim of increasing voter turnout and registration amongst sectors where the likelihood of backing their candidate was high, and influencing voter choice in sectors where the support metric influenced voters could go either way.
This meant that targeted messages could be despatched – via email, social media posts and browser display ads – depending on whether an individual needed to be convinced to register, vote, or pick the correct candidate.
Doing this is relatively simple – you simply look at what has worked in the past. For example, the activities involving increasing turnout focused on how individuals had previously been encouraged, and whether this had impacted the probability of them leaving the house to vote on election day. One study recently indicated that very simple cues such as informing people who have said they will vote that they will be re-contacted after the election to see if they actually did can have a measurable influence. Do this on a massive scale involving millions of people across a nation and you have a Big Data election campaign.
In the years since then, all parties and candidates – with the one notable exception discussed previously – have enthusiastically launched their own analytics strategies. Platforms such as NGP VAN’s Votebuilder are used by the Democratic Party and their analyses are made available to all candidates. The Republican national congress in response launched their own startup – Para Bellum Labs – to assist their candidates.
On an individual level, too, candidates (except Trump) have publicly entered into partnerships and arrangements with data strategists and lent their endorsement to political analytics platforms.
Running against Clinton or Sanders in November, this could mean that Trump finds himself at a distinct disadvantage. Both Democrat candidates’ campaigns are backed by technology spun out of that which was developed by Obama’s groundbreaking analytics teams. And those technologies have not stood still. Now commercially packaged and operated by independent consulting and vending entities, they have spent the last four years absorbing cutting edge advances such as machine learning and natural language processing into the analytical mix.
Eschewing (although the extent to which he has done this is, as noted before, debatable) solicitation of private donations may bolster Trump’s populist edge, but it also deprives him of a valuable source of data. Campaign contributors are largely happy to share their details. This allows researchers a peek at their socioeconomic status and contributes to the profile which is used to identify other potentially friendly, but undecided voters. Either Clinton or Sanders is likely to have far more of this information at their fingertips when it comes to the last leg of the contest for the presidency thanks to more comprehensive efforts to build voter and donator databases.
Whether this will prove enough to counter the undeniably unprecedented surge of popular support for the property magnate turned reality TV star remains to be seen. Political commentators have identified Trump’s ability to resonate with disenfranchised Republicans unhappy with the direction the party has taken in recent years as key to his success. But even this nebulous demographic – less likely to be engaged with the political system in general, but keen to have their voices heard – is likely to have divulged enough personal data to give Trump’s opponent, whoever that might be, a shot at changing their minds. One thing that is certain is that both Big Data advocates and those who are critical of its ability to implement true change will be following the 2016 election with bated breath.
