Donald Trump won the presidency using the social media ad mechanism exactly as the company intended. He is ready to start again.
Look at the millions of Facebook ads posted by Donald Trump. It’s hard to believe they’re a winning strategy.
The Trump team recycles the same imagery and the same themes over and over again. Here is what we see most often:
- Trump, photographed in front of an American flag, and pointing;
- Trump applauds in front of an audience;
- Trump gives a thumbs up or smiles at a microphone.
Each image is denoted in patriotic red or blue.
The text almost always calls for action: buy the “Make America great again” cap, sign this petition, go to this gathering.
A race for Facebook advertising
Trump’s Facebook ads are remarkable for their banality and volume.
During the 2016 election cycle, the Trump team released 5.9 million Facebook ads, spending $ 44 million June to November only.
His opponent, Hillary Clinton, also bet on Facebook, but on a smaller scale.
Hillary Clinton’s team appears to have launched 66,000 Facebook ads. We are therefore very far from the 6 million advertisements of his rival Donald Trump.
Meanwhile, Bernie Sanders bought only 8,400 ads, Elizabeth Warren and Joe Biden even less.
Everyone uses Facebook, but Trump uses it in a different and visibly better way.
Earlier this year, Andrew “Boz” Bosworth, who led Facebook’s advertising team in the 2016 election, wrote that Trump “had run the best digital advertising campaign I’ve ever seen from an advertiser” . The Trump team agrees, of course.
Trump has not mastered Facebook due to foreign interference from Russia or psychographic exploitation via Cambridge Analytica.
It did not do so via micro-targeting – the ability to send the right messages to very different audiences to change attitudes or inspire action.
In reality, Trump used Facebook exactly as Facebook intended: to pay insane amounts and accumulate a lot of data in Facebook’s automated advertising system.
Facebook connected a machine to electoral decision-making. Political campaigns stopped communicating with voters and began to communicate with AI instead.
Facebook’s artificial intelligence is a crucial element for a winning campaign in 2020.
Profound changes in advertising activity
In the past, advertisers bought guaranteed locations in print publications, on outdoor screens, or in media shows.
They chose these placements depending on the audience that these media could reach: a glossy magazine for women interested in fashion or a billboard seen by thousands of people on their commute.
At the dawn of the internet, advertisers did the same thing on websites. If a company wants to position itself in front of a particular audience, it can buy space on a site attracting this audience.
Facebook turned this upside down. A “Facebook ad” is less an ad and more a machine for producing ads.
Instead of paying to put particular media in front of a specific audience, an advertiser now pays Facebook for provide a selected result on a category of people.
For example, a clothing manufacturer can pay Facebook for visits to its site from women in their thirties who live in Los Angeles.
Therefore, the way in which these advertisements reach the corresponding users depends on Facebook.
Given certain information at the outset, its system learns to regulate the distribution of advertising compared to all other advertisers.
In short, Facebook chooses which ads will be shown to whom and at what price.
Today’s advertisers collect the raw information (real and potential customer data) that allows Facebook to do the job on their behalf.
What you need to understand is that when these ads succeed in getting users to act, these actions always generate more data.
This feeds the Facebook system which has more data to analyze, and which can optimize the dissemination of your advertisements.
A well-oiled gear
Advertising previously involved identifying a market for products and services and then placing media to respond to that market.
Now, that means train artificial intelligence to gather the right audience from actions performed by smaller samples of people.
Most political ads on Facebook start with user lists, which Facebook calls “custom audiences.”
A custom audience can be created from data that an advertiser has collected or acquired outside of Facebook – lists of phone numbers or email addresses, for example.
A custom audience can also consist of anonymized Facebook IDs generated when users perform an action on a web page.
Of course, it is essential to have installed the Facebook pixel beforehand, a means of tracking online activity to its source.
Once advertisers have identified their audience, they create “lookalikes” – groups of people who Facebook says will act the same as those included in a personalized audience.
When you find a successful advertising process – more people signing up for an event, buying a cap, or giving a phone number – you feed the Facebook system.
If it works properly, the system can develop naturally: more money means more ads. So you’ll want to invest more money and create more ads.
This is great news for Facebook because once you start buying ads there, it becomes difficult to stop.
The millions of dollars spent by Trump
Trump spends tens of millions of dollars on Facebook marketing. These annoying ads work so well because he decided to cede control to Facebook’s advertising purchasing mechanisms.
With all the data collected, it’s easier to get even more results.
Facebook wants users to spend as much time as possible on its site. To encourage this, the company is making more attractive ads less expensive to launch.
To cut costs, marketers can create better performing ads or find audiences that respond well to ads already created in the past.
The Trump team bet on the latter.
Remember, however, that a Facebook “ad” isn’t just a picture and a link.
An advertisement is actually the combination of content and audiences that advertisers want to push into action.
For years, Facebook has helped advertisers achieve this goal at the lowest price.
The company is now encouraging advertisers to hand over the reins entirely, letting Facebook divide spending among ads, audiences, schedules and budgets.
The company even automates auctions on behalf of the advertiser.
As the Marketing Addict put it, “Facebook’s advertising algorithm has become so much better at automating campaign management that it can now easily outperform a human manager.”
This is why Facebook advises advertisers to “give freedom to placement or audience algorithms. This gives systems more opportunities to consider when evaluating which one will perform best. ”
To turn the crank that starts this machine, marketers must develop their own knowledge of users that they have or want to reach.
The best way to do this is to create custom audiences. Your audience must be large and qualified. From then on, your results can be interesting.
Initially, the Trump team allocated very little money to a small number of ad variations.
Then the Facebook machine took over, trying to maximize the “value” of advertising in the middle of friend posts (what Facebook calls “organic content”).
Ads based on AI predictions
As with all machine learning systems, the results can be difficult to relate to the causes.
When Facebook shows you a particular ad, it does so on the basis ofa projection of you, consisting of actions you have (or have not taken) on the content that Facebook follows.
The elements you click on make up part of this signal. The way other people with similar preferences have interacted with similar ads does too.
Facebook is now making its predictions on 2 million “features” separate, said Goldman.
This could be the last time someone saw an ad ate a burger, the minute an ad was launched, or the percentage of battery life remaining on someone’s phone.
If there are multiple ways to collect data with Facebook, machine learning models have tried to account for it.
Do predictions make a good model of a person’s real wishes? Do the ads work? Whatever.
Facebook adware is not trying to get someone to buy a product or vote for a candidate.
He’s just trying to produce the results advertisers say they want, serving ads to users similar to those who provided these results on previous similar ads.
Every action a user takes or doesn’t take – clicking, liking, sharing, commenting, giving, hovering, buying, filling out a form – slightly alters the complex network of predictions that make up a person’s Facebook image, c is to say: a consumer.
From Facebook’s perspective, big political advertisers are just additional data sources to integrate and optimize for the business, not competitors vying to rule the free world.
The place of algorithms in advertising activity
The new world of buying ads automatically by machine learning launches a loop for human marketers.
The algorithm provides feedback in the form of campaign results (site visits, registrations, etc.), but no one really knows which signals will serve ads at a fairly constant price and volume.
Trying the same ad tomorrow could produce different results than today.
Over time, Facebook has exercised more and more control over the advertisements displayed, to whom and for how much money.
Paradoxically, this means that marketers must constantly monitor what Facebook is doing, lest they end up spending too much or too little.
What you need to do is learn to communicate with incredibly complex artificial intelligence, so that it provides ads that produce the results you think you want.
Trump never stopped advertising
According to a source familiar with the Trump 2016 campaign, his advantage came from a surprising inspiration: the mobile game company Machine Zone.
Machine Zone was one of the first to display high-volume Facebook ads.
At one point, the company said it was “probably the greatest direct response marketer in the world,” thanks to its ubiquitous ads for games like Mobile Strike and Game of War.
Machine Zone had even built a set of tools that allowed it to do this type of audience building and buying Facebook ads before Facebook and other service providers allowed it.
Machine Zone does not appear to have worked directly on the Trump campaign, but according to a former Machine Zone employee, his methods have inspired Trump’s digital team to adopt this type of advertising strategy.
From there, the Trump team created a small Facebook fundraising machine.
It didn’t take as much technical genius as the help of people like Coby on the Republican National Committee, James Barnes, a Trump campaign Facebook worker, and their worker bees in ad buying companies.
Together, they made modern elections entirely cybernetic, a complex and dynamic interweaving of man and machine that is totally different from what came before.
As we approach the 2020 general election, Trump has a new form of historic advantage. People are amazed that Trump has never stopped running Facebook advertising campaigns.
And the reason is thathe could not. The point is, the campaign must keep new data flowing through the system.
Most of the time, it can optimize the cost of acquisitions, by collecting money and data from the Facebook users it targets.
Then, at strategic moments, the team overturns the machine, spending all the money necessary to achieve the highest and widest reach among its employees.
COVID-19 pandemic changes Trump’s Facebook campaign, but it also created new opportunities for its Democratic competitors.
By mid-March, as the new coronavirus began to shut down America, the cost of Facebook ad placements had dropped almost 40%, according to Kazim, the founder of Facebook-marketing-firm.
Indeed, although more people are spending more time on Facebook searching for information and comfort, more advertisers have cut back on spending due to the economic calamity.
He threw a key into Facebook’s advertising machine, offering an advantage to advertisers who pay attention to it.
A machine-driven future?
Trump’s 2016 campaign never really ended, and having so much data in and out of Facebook probably gave him an edge.
It is a problem.
Facebook prides itself on the supposed transparency of its political ad archive, but this resource provides insufficient detail.
If policymakers or election officials wanted to play political advertising on Facebook, they could ban the use of personalized and similar audiences for campaigns.
As Facebook’s profits increase, the machine learning ad serving system continues to grow, quietly and perpetually.
From time to time, glimpses of its massive reach spring from the dark.
Earlier this year, Facebook announced the launch of its non-Facebook activity management tool, which allows users to see and control how organizations like the Trump campaign can use information collected outside of Facebook (from ” and voter registration), and then retarget them for example.
New controls allow users not to allow Facebook to use non-Facebook activities to target them, but that doesn’t stop organizations like the Trump campaign from sending this data to Facebook.
This does not prevent Facebook to use the data it receives to train and operate its machine learning ad targeting systems
These continuously improve and become more and more incomprehensible to humans.
Marketing on Facebook is not so much an advertising strategy as a new lifestyle led in the midst of an invisible extraterrestrial intelligence.
Baffled but willing, advertisers like Kazim embrace their extraterrestrial impresario, in part because they have no other choice: “I think letting Facebook do its job is the future,” he said. .
They are not alone. The users – who are also citizens – also have no way out.
Letting Facebook do its job has become a requirement for electoral politics, and the future of democracy is tied to results.
To read the full article in its original version, you can follow this link.