Poll Position: Ad Industry Casts Its Vote on US Election Polling

 

Election season is all about the polls, and marketing presentations are increasingly all about data. There is something about numbers that lends authority to any argument, a natural tendency to trust statistics, as if a scientific theorem has been proven versus an argument made.

 

But as Mark Twain once said, “There are three kinds of lies: lies, damned lies, and statistics.” Data is incredibly useful, but savvy marketers know data is only as good as its source, and more importantly, only one input into a messy process.

 

Without supplementing survey data with one-on-one conversations to gauge enthusiasm and intent, pollsters and marketers are looking at a blurry snapshot. The nature of human behaviour makes polling a pale proxy for what’s going to really happen. Your opinionated neighbours next door with the sign in their front yard may enthusiastically answer the call from a pollster, but someone less extroverted may ignore the call or delete the text.

 

That’s why pundits admit that lopsided polls are sometimes used to influence behaviour – why bother to vote if it looks like your candidate will get crushed or win by a landslide? Similarly, some marketers use data to influence decisions from upper management and justify their budgets rather than invest in research that shapes campaign development.

 

The smart approach lies in finding the right balance between quantitative data, qualitative interviews, and a healthy dose of common sense. Data is a signpost but not a crystal ball, so use it wisely.

Marketing analysts and pollsters face a lot of the same challenges – how can we collect self-reported data that reflects real-world behaviour?  How do we balance respondent profiles to fit the category we’re measuring?  

 

Perhaps the most material shared challenge is studying the people who are most important to growth or success. In advertising this tends to be the light category buyers that are open to switching brands and in polling this is the true undecideds who are often lower information voters and have less predictable voting patterns.

 

The biggest difference between marketing and political polling is how the outcomes are distributed. In marketing, econometric models predict unit sales or shifts in market share. Small deviations from the forecast tend not to make or break a brand’s fortune. In politics, all of the electoral votes of a given state are allotted to the winner of the popular vote in that state regardless of margin. There was a lot of discussion in the last election cycle about pollsters getting it wrong, and the belief that it was due to people not wanting to share their perspective.  Post analysis showed that the major culprit in certain states was that the difference in the election between the two candidates was so small, it fell within the sample error range.  

 

Marketing analysts, and I’m sure brands as well, are thrilled to not live with such a binary outcome. 

 

While political polling may present a lot of challenges, there’s one thing political campaign teams have going for them that anyone in adland should be envious of, and that’s putting data at the forefront of a team’s operations. Predicting the future is never going to be easy, but polls still do an overwhelmingly good job of what any data-based strategy should be doing – helping campaigns zero in on their biggest challenges and opportunities, just look at the focus on swing states in the US presidential election. The closeness of the race in those states is omnipresent for everyone working on either campaign, guiding their actions, and providing a clear sense of urgency. With such a clear goal, strategic and creative thinking can enter the chat. 

 

Beyond the obvious of aligning on the role of data and measurement for any organisation, polling highlights the benefits of investing in high quality consumer research over time for any business. Remember that a lot of the polls conducted for elections are done in collaboration with universities and news organisations – all to say that even small businesses often have the resources to invest in some level of higher fidelity research and analysis. In business, the clients that do this best have the benefit of being able to parse out proactive long-term strategies from purely short-term reactive thinking. Anyone in adland should be looking at ‘polls’ and asking themselves, how can more consistent, real-time, and quick turn insights help us avoid such a close race with the competition?

 

The political universe, and those connected to it, provides a masterclass in creating narratives. There is no industry or entity better at spinning data into headlines. Not a week goes by without a story on favorability ratings, Americans’ views on major issues, or who is leading in a local, state or national race. Politics knows that facts don’t travel, stories do. But those stories need one little piece of evidence to help them go viral. Research groups, think tanks, news organisations, and PACs know how to play this game well – use a data point to create a story that affirms one position and infuriates the other. Sit back and watch it spread like wildfire. 

 

While there are countless ‘predictions’ in political data, the end goal isn’t about being accurate. It’s about story. Think about it. No matter who wins this presidential election, the headlines will be about whether the polls were right or wrong, and which ones got it ‘right’. The errors in sampling, weighting, survey bias, and selective analysis are well documented in politics. And in coin-flip elections, even the most accurate of polls, with a 3-6% margin of error, are challenged to get it right. Win or lose, there will be a story. That’s what matters. 

 

Cynicism aside, it is an impressive display of how to use data to tell a story. It’s a case study for the marketing world. Marketers are obsessed with using data to optimise campaigns, affirm pre-existing viewpoints, and to de-risk decision making. The major difference is marketers look to data to provide certainty – to predict that X action will lead to Y result. 

 

Perhaps this is the wrong way to view data, not least of which because it’s impossible to predict the future. Almost every political poll got the 2016 election wrong. That hasn’t slowed down the amount of polling, the media’s obsession with cherry picking one poll over another, or Americans' willingness to believe them when it fits their desired narrative. 

 

Rather than looking at data as the single source of truth, marketers should take a page from politics and look at data as the main character in a bigger story.

 

People’s obsession with political polling always makes me think of a classic quote about market research. ‘People don’t think what they feel, they don’t say what they think, and they don’t do what they say.’ Polling is just good old-fashioned market research, knocking on doors, asking people what they think, helpful for getting insight to inspire marketing, but not always correlating with real-world behaviour. In uncertain times, we crave the security of knowing what happens next but that uncertainty makes it harder than ever to predict people’s behaviour, especially when said behaviour might be considered controversial. That’s why different pollsters predict everything from the closest-ever election to political landslides. And that’s why, as marketers, we must use ‘polling’ judiciously, getting beyond the rational (are you persuaded?) to the emotional (how do you feel?), which helps us understand people and de-risks investments but isn’t the ultimate arbiter of truth.  Our focus should be on behavioural data, testing with and learning from real-world actions.  Which, in the case of the election, we’ll only have once people get out and vote.  

Image Credit Katelyn Perry Unsplash+