"AdExchanger Politics" is a weekly column tracking developments in the 2016 political campaign cycle.
Today’s column is written by Peter Pasi, vice president of political sales at Collective.
Borrell Associates predicts candidates will spend $1 billion on political digital advertising ahead of the 2016 election. While that’s only a fraction of the total advertising picture, it’s a huge jump from the $159 million spent on digital outreach in 2012.
Political programmatic advertising requires a dramatically different skill set than more traditional brand advertising, particularly where data is concerned. That’s largely because political advertisers face far more challenges that brand marketers when using first-party data and matching it to voters. That’s forced many to dig deeper to find likely voters.
Marketers use first-party data to target consumers and persuade them to buy laundry detergent or a subcompact car, but almost any adult consumer will be in the market for these products at some point in his or her life. However, in political advertising, first-party data limits the size of an audience rather than helps expand it. This is because only qualified voters can vote in elections; only registered Democrats, for instance, can vote in closed Democratic primaries. As a result, all political advertising needs to be highly targeted to reach a very specific first-party audience.
Political audiences also fall into extremes, and those diametrically opposed audiences are known and finite. Yet only a minority of voters can be found via first-party data matching. This forces political marketers to rely heavily on modeling, which has its shortcomings. Since an impression delivered to the wrong person is an impression wasted, political marketers must find ways to leverage data wisely.
Ultimately, an election is decided by a relatively small number of voters. In presidential elections, only half of eligible voters actually show up – and in many cases those voters aren’t “persuadable.” In 2012, Rick Santorum won the Iowa caucuses by 34 votes. That’s a lot of pressure on a marketer to target, reach and persuade a very small number of voters. It’s very different than trying to persuade them to buy a can of ginger ale because there’s simply more at stake. And in most electoral contests, there’s no prize for second place.
The Rise Of Microtargeting
Political marketers are therefore forced to consider a variety of approaches to reach undecided voters. George W. Bush was the first president to use modern microtargeting, digging beyond known voter data to glean insights about the electorate. Microtargeting is rising in importance in the 2016 election, when creativity is critical and first-party data and likely voter models are becoming less predictive. Even when the party of the voter is known, the likelihood of that voter turning out to vote can sometimes be inaccurately predicted.
In the past, simple microtargeting models looked at race, age, income and a handful of other variables to try to predict voting behavior. But in the past decade, the sheer amount of data available to campaigns, combined with significant improvements in computing power and a renewed interest and focus on predictive models, has enabled campaigns to unearth new prospective supporters and conduct targeted outreach to those voters.
Today it’s possible to take thousands of pieces of research and data about a group of voters and process them with software in the cloud or on a laptop; these processes may have required a mainframe 20 years ago. And results of polling, ID calls, field work, email and direct mail responses and TV viewership gleaned from set-top box data can be input and processed in near-real time. This unprecedented level of data analysis allows campaigns to respond to changes in the political environment and refine their persuasion and turnout models on a moment’s notice, which can mean the difference between triumph and defeat.
Political marketers need to look beyond first-party data at facts and purchase behavior that may indicate voters’ political leanings. For example, drivers who own hybrid vehicles tend to be Democrats, but that’s not strictly the case since there are certainly Republicans who like to save money on gas. However, if a voter drives a Tesla and also shops at LL Bean or Whole Foods, they’re likely to be a Democrat. (This Time magazine chart offers some interesting insights on shopping behavior along party lines.)
A political marketer can identify voters they believe to be sympathetic to their candidate and isolate characteristics those voters share at levels that exceed those of the general population. For example, if a marketer notices that voters who support her candidate are in-market for a Ford, use a Discover card and like frozen yogurt, she can find others who might share those behaviors and might also be likely supporters. By adding digital data to offline data, that marketer can make stronger connections and increase the pool of potential supporters.
Responding To Volatility
Political campaign marketers have historically relied on first-party data to try to understand who voters are and what matters to them. However, static voter data alone cannot be relied upon to win an election because there are always surprises. Districts that are historically Democratic can unexpectedly vote Republican because a candidate doesn’t conform on a single issue or because he or she behaved badly at a fundraiser. A polarizing candidate can turn voter expectations upside-down. Candidates are becoming less like the proverbial cookie cutter so campaigns must follow suit.
Marketers need to continuously optimize their messages, data and media in a volatile, fast-paced political campaign. There’s more on the line for them than for that laundry detergent or soft drink marketer. There are no second chances.
Behavioral and purchase history data can also support these efforts: We can glean a lot about voters based on what they read, watch, share and buy. Political marketers also know that different issues matter to different demographics. For example, data can easily help marketers identify new mothers in the digital world. Swing voters in this demographic may be more sympathetic to a candidate with a strong platform on child care or child tax credits. Using the same strategies, a hunter will also be relatively easy to find online and may have strong views on gun control and related issues. The data can reveal a great deal and provide political marketers with some signs to guide them along the campaign trail to victory.
Political campaign marketers have to be smart, savvy and persistent. Elections are a tricky, emotional and unpredictable business. However, data in the digital medium gives marketers insights that have never before been available to them. With access to deeper information about their voter audiences, campaign marketers have the opportunity to be more targeted than ever, and perhaps even smarter.