“Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Martin Kihn, research vice president at Gartner.
People who believe Cambridge Analytica put our president in office aren't thinking very hard.
We know how advertising works: It hardly works at all. Even perfect insights can't survive collision with the forces of ad ops, dynamic creative, multichannel measurement and – worst of all – the human mind.
Cambridge Analytica is a consultancy accused of many things, from genius to crime. We'll leave the ethical and legal issues to those more qualified to handle them and focus on an assumption that lurks, unspoken, behind most of the current hysteria about Facebook, Russians, 2016 and beyond.
A heart-warming number of people assume that advertising has magical powers. Those who know it better only wish they were right.
So, in the spirit of transparency, I offer you three myths about advertising:
Myth 1: Advertising has magical powers
People ascribe great impact to a medium they increasingly ignore.
In times of crisis, language about advertising becomes almost cosmic. Whistleblower Christopher Wylie claimed Cambridge Analytica’s goal was to "exploit what we knew" about the 50 million Facebook profiles it had and "target their inner demons."
Demons! We've been here before. In the late 1950s, Vance Packard shocked the world with “The Hidden Persuaders” and singled out the Subliminal Projection Co. for flashing "Drink Coca-Cola" and "Hungry? Eat Popcorn!" messages at a drive-in movie theater in New Jersey. Sales soared.
Media were outraged. Said the Saturday Review: "The subconscious mind is the most delicate part of the most delicate apparatus in the entire universe. It is not to be smudged, sullied or twisted in order to boost the sales of popcorn."
Amen to that, of course, but wait: The Subliminal Projection Co.’s founder later admitted he'd made up the experiment. Legislation was loudly introduced but went nowhere. There were no hearings.
Today's subliminal Satan is “psychological profiling.” Cambridge Analytica’s approach was apparently to analyze what people "liked" on Facebook to construct profiles. Inspired by academic studies with Orwellian titles such as "Psychological targeting as an effective approach to digital mass persuasion," it used likes to calibrate personality types and then craft ads.
But there is a whole lot of road between labeling someone an introvert and controlling their brain. (Some obvious problems are outlined here.) In fact, there is plenty of evidence that Cambridge Analytica made mistakes and underwhelmed its advertising clients. Magical, it's not.
Myth 2: Advertising wins elections
Advertising is supposed to have some influence. That's the point. So, unless we're willing to say that all ad spend is just shredded lettuce, we should expect most campaigns to do something.
But what? It's interesting to discover that the case for advertising's impact on politics is so weak. It's much harder to influence a vote than a snacking decision.
As it turns out, there are hundreds of academic studies on this topic. Many use a method familiar to mix modelers: combining media weight and survey response data at the market level. Applied to recent US presidential races, these models found either little impact or no impact, aka poor ROI.
Ballots are secret, of course, so academics study proxy metrics like seasoned brand marketers. They examine upper-funnel (favorability) and lower-funnel (get out the vote) KPIs.
Upper-funnel impacts are all over the place. Farther down the voter journey, things are worse. One expert summary of the impact of ads on voter turnout concludes: "The consensus view leans in the direction of no effect."
Alas. Interestingly, one study of US Senate races found that certain things made ads work better:
- Being an unknown
- Being a challenger
Our recent presidential race had a challenger. Yet there's evidence that ads had even less impact in this election than usual, in part because of massive unpaid – earned – media coverage.
Myth 3: Advertising works
OK, I’m kidding here. It works, right? Right?
Those of us who have spent time trying to measure the impact of, say, brand campaigns can be excused for our cynicism. There is plenty of evidence that ads have an incremental impact on a potato chip purchase, but there are a lot of moving parts in a campaign.
Try this thought experiment. Imagine you are a mysterious British consulting firm with 50 million names and a list of things they liked, including some guy named Martin who likes, oh, Bernese mountain dogs, Taylor Swift, CrossFit and ballet. You have this amazing psychological model. You want to change his vote.
To succeed, you must:
- Have accurate data (people lie all the time)
- Have current data (maybe he's switched to dachshunds … as if)
- Have enough data (looks pretty thin)
- Assume he knows what he likes (apparently, we really don't)
- Find a match while onboarding (rates are 30-40%)
- Find 999 people just like him (there are minimum audience sizes)
- Bid efficiently against competitors (unless money is no object)
At this point, you've simply negotiated programmatic platforms. Now you're up against the realities of all advertising, which works (or does not) on the human brain. In order to work, ads need both emotional content (to be noticed) and frequent exposure (to be remembered). So, you need a lot of media weight, particularly around Election Day.
Finally, you need the most important element of all: good creative.
Creativity matters more than we think. It's the most important single element of anything but the twitchiest coupon ad. Nielsen Catalina Solutions showed that creative elements accounted for about half of a CPG campaign’s impact, on average.
Here is where we find the real magic. Whatever your opinion of its methods, Cambridge Analytica did have a secret weapon – just not the one we’ve been talking about. After the recent scandal broke, Cambridge Analytica’s CEO boasted about "weaponizing" the slogan "Defeat Crooked Hillary."
It's ironic that the poster boy for data-driven marketing may have succeeded because his agency excelled at something else: They are promising copywriters.
Before I ignite the message boards, let me be clear: Some Cambridge Analytica ads probably had some impact on some people. In any campaign, some ads have some impact. It's a question of proportion. Let us never confuse advertising with the fundamental forces that should really trouble us.