As originally featured in AdNews.

This year’s election has broken so many rules that it barely resembles a political process. Few predicted that Donald Trump would be the last Republican candidate standing, let alone that the Grand Old Party would subsequently rally around him.

Setting politics aside, any marketer has to admire his intuitive read on the American electorate. Trump recognised a deep unhappiness that the pollsters and politicians somehow missed.

Part of Trump’s strategy has been to ignore the data that powers all our marketing campaigns. It may seem foolish to ignore the power of data, but let’s not fall into the trap of underestimating The Donald – he is redefining what it means to be a Republican voter.

But what can data tell us about Trump supporters and the characteristics that define them? And should Trump evolve his strategy as we approach the general vote?

At Exponential, we used data to analyse the profiles of traditional Republican supporters and compared them to those likely to vote for Trump. Interestingly, of the 13,059 statistically significant characteristics that define a Trump voter, only 40.1% share the direction and strength of Republican loyalists. It’s enough to make Trump a Republican, but not enough to have most Republicans support Trump.

Here is a comparison of some traits between Trump voters and Republicans.

So how should we understand a Trump voter? Two key themes dominate this profile: uncertainty and pride.

Uncertainty

The strongest (and most sobering) pattern for Trump loyalists is a pervasive sense of uncertainty. It is a stark contrast from the lives of the traditional Republican voters, who are well-established and comfortable.

Trump appeals to those struggling with economic uncertainty. His core supporter has poor credit, earns less than $15,000 a year and owns a home worth less than $50,000. Trump voters are 52% more likely to be unemployed than the average American. Understandably, they postpone most discretionary purchases and frequent off-price and dollar stores, as well as Wal-Mart.

Their sense of uncertainty extends well past income. Trump voters are 31% more likely to need car repairs, 2.4 times more likely to suffer from insomnia, and three times more likely to seek marriage counseling. Stress is evidenced by high rates of drinking, smoking and use of diet pills.

None of these patterns are evident in traditional Republican profiles. These people earn more, have larger homes, drive more expensive cars, and still manage to save for retirement. Traditional Republicans can be found shopping for jewelry, cosmetics, and luggage – behaviours that almost don’t exist in the Trump camp. Loyalists are 3.7 times more likely to consider adding a swimming pool, an investment that can cost as much as a Trump supporter’s annual income.

Pride

For Trump supporters American culture is prized, and this is reflected in favour of American cuisine, especially a barbeque. They read about American history and visit American historical sites. They prefer cars with American brand names, like Jeep and Ford. And they listen to country music.

Cultural interests offer an easy shortcut to find Trump supporters. For example, Zac Brown Band fans are 1.9 times more likely to have voted for him, as are viewers of Two and A Half Men or The Bachelor. Negative targeting is just as useful. Infrequent behaviours include interest in imported cars such as Audi or Acura, or music preferences in hip-hop or K-pop.

Trump voters are two times more likely to drink beer than Republican loyalists, who conversely are two times more likely to drink wine than Trump core voters. Trumpsters follow football, while Republicans prefer baseball. And perhaps most tellingly, Trump voters tend to live alongside significant minority populations, while loyalists live in areas that are overwhelmingly Caucasian.

We are already seeing these voter profiles change now that the primaries are over. Trump supporters today, growing in number as he brings in reluctant Republicans who had favored Jeb or Marco, make his audience look more and more like traditional Republicans. Does this mean Trump needs to change his views on data?

I would argue that in the general election, data remains as important as ever. As successful as Trump has been in commanding attention and engendering passionate support from his core audience, he’s been rallying the opposition as well. Obama won his elections by identifying and activating Democrats in Republican areas without awakening the opposition – action without reaction, if you will.

Instinct may have gotten Trump this far, but the latest polls suggest he still has a lot of ground to cover if he intends on winning the election.

At the very least, this will involve a sustained get-out-the-vote effort that targets individuals with custom advertising messages based on their known preferences and characteristics. He is sure to stumble without the deep knowledge and precise targeting that data can provide.

How we did it

To determine these profiles, Exponential profiled residents of US counties where Trump led the primary vote by more than 20 points. This study covered roughly 100 counties and 18,248,560 user devices on the Exponential network. To validate the findings, Exponential also modeled the preferences of Caucasian males in the United States that had a high-school education and were working in agriculture, labor, or unskilled positions. About 71,080 users were considered in this approach.