The day before the recent US election, the New York Times ran an article with the byline ‘Several factors that led to the misfire last time are no longer in play’. It argued, backed by quantifiable data, that 2020 would be different.
Of course, it wasn’t. A 10% Biden lead in Wisconsin ended up as a 0.7% victory, and a 2% lead in Florida turned out to be a 3.3% loss. And Rep Senator Susan Collins who did not lead a single external poll easily won in Maine.
But we shouldn’t be surprised pollsters got it so wrong.
The polling industry has been heading on a path to self-destruction for years.
A path that will sound familiar to anyone who’s been in advertising for a while.
Around 2008, US political journalists became obsessed with data, turning the likes of Nate Silver into data pop-culture celebs. Even though the sample sizes were often small, data became a story in itself as TV pundits practically orgasmed over one or two percent swings.
And as the data percentages produced more headlines, the stories that could have explained what was really going on in American culture were rarely told or deemed irrelevant.
There was lots of quantifiable data, but less and less understanding of the underlying context. Somehow voters were persuaded to believe that the polling errors of 2016 were an anomaly. But 2020 is ground zero.
While this is undeniably terrible for the polling industry, it’s a great opportunity for the advertising industry to reassess our approach to understanding culture.
In the last decade, we’ve all watched advertising’s handsy teenage romance with quantifiable data. Like pollsters, advertisers' view of this type of data quickly morphed from being a new and useful source of information into an obsession.
Despite this, most advertising still doesn’t permeate culture in any meaningful way.
And that’s a huge problem given that culture is a much more powerful force than marketing can ever hope to be.
As the election showed, data is only useful if you truly understand its context.
Part of the issue is people often lie, have secrets or simply don’t trust complete strangers enough to say what they really think. Most of us have become increasingly aware that once we give out information, we can’t control where it ends up. What if, somehow, it comes back to bite us?
Applying ‘gut instinct’ to the information we learn isn’t the answer to advertising’s dilemma either. Interpretation of data is always subject to our own bias and the limitations of our
own worldview.
This is startlingly clear in the debate that’s raging between moderate and progressive democrats on the roles of BLM, Green New Deal and Defund the Police.
Each group is cherry picking stats to prove these causes either harmed or helped the Democratic vote, depending on their point of view.
Unfortunately human nature dictates that we often hope data proves us right. And when it seems off, we tend to ignore it or simply twist it to draw the wrong conclusion or a conclusion which we’d already predetermined.
The situation is compounded by categorising people into homogeneous blocks that lack any nuance whatsoever. Let’s take the term ‘Hispanic’, a ridiculous catch-all grouping that ignores a diversity of beliefs, background, interests and cultures. By the logic of polling, Ted Cruz and Alexandria Ocasio-Cortez should have a similar political outlook. It was also interesting to watch the confusion among commentators as top 1% net worth individuals Lil Wayne and 50 Cent expressed some admiration for Trump.
These neat little blocks of bad demographic assumptions have been the punchline in many an advertising brief. Millennials, Gen Z? So whole generations have the same values, backgrounds and behavior? Unlikely.
And then there are little gems like this; “She is driven by her passions, her career and her family - but she doesn’t expect everything to be perfect”. She’s the (insert catchy descriptor).
It’s a wonder advertising resonates with anyone at all.
So, what can we do?
More and more agencies, ours among them, believe in a more radical ethnographic approach. We want people to teach us about their lives and feelings so we understand the nuances of why they, and others around them, behave the way they do.
Of course, ethnographic research isn’t new.
But the real opportunity lies in making lateral decisions about the people we want to learn from and those who will get them to open up to reveal unheard truths.
For example it’s unlikely many agencies would include sex workers as part of their research and if they did, it would be to talk about sex.
Chances are they could teach us lots about many topics thanks to the multi-faceted nature of their job. But it goes without saying that trust between both parties is paramount and that’s not an easy thing to achieve.
People like journalists, whose job is to unearth secrets, have the skill sets to provide us with the context we sorely need. And yes, what we’d learn would be richer and incredibly more nuanced than a sheet of data.
During the election some journalists from non US publications talked to people in small, ignored communities to learn about their lives, feelings, hopes and fears rather than how they were going to vote.
The pictures they painted were of rage, despair and distrust of politicians.
And guess who they described as ‘Not a politician’? Yep, you got it.
Of course, no one expects quantifiable data to go away and nor should it. But what should disappear is blind faith in it as the main (and absolute) predictor of human behavior. We need much more nuanced understanding of the context.
The elections, the only poll that counts, have proven that.
So, the next time someone proclaims videos must have a brand mention in the first three seconds, you might not want to listen.
Maybe, just maybe, 3 seconds is the time it takes for it to be registered as a view before you can scroll past it. Not that you can’t prove it, of course.
Michael Murray is founding partner (creative) of Berners Bowie Lee