Programmatic advertising, the Internet savvy and AI

Let's start by explaining a concept that is well known in digital marketing, but not so much in public forums, the "programmatic advertising". This type of digital advertising is what you usually see every time you visit an online publication or many types of informational websites.

It works as followsAdvertising: Companies that advertise on the Internet do so by automatically bidding for available advertising space on a network that may be managed by Google, Meta or another provider of such advertising services. Google Ads is the undisputed leader in this field.154 billion, with approximately advertising revenues in 2023.

When an advertiser designs his campaign, he decides on a series of parameters on the target you want to go to.

As the advertiser is going to pay for the number of times his ad is shown (or the clicks on it) he needs to parameterize it in detail, so that the advertisements are shown to the type of customer you want to reach and not to others, maximizing what is called cost per click or per views.

The more data we have on the targetThe easier it will be for the advertiser to design an ad just for that subject.

The parameters to be chosen vary according to the platform, but basically you decide on geographic aspects, purchasing power, interests, and specific days and times when you want the ad to appear. This is where the algorithms of each platform do their work. magicand for which both Google and Meta have frequently been accused of collecting too much data from Internet users, in most cases without the users' knowledge, but that's another story.

Advertisers do not know where their ads end up being published.

The result of this sum, parameterization plus algorithmsThis is what results in advertisers paying for their carefully designed campaigns to end up on websites they don't even know exist, since almost no one personally monitors where the ads actually appear.

Introducing the bad guy: the websites "created for advertising".

A well-known and common practice on the Internet that takes advantage of this system, is the creation of targeted web sites for the advertisements of the advertising networks to be published on the site from Google. These websites register in the system, and Google starts sending ads to them. These publications earn revenue from the publication of those ads there, a percentage of what Google charges the advertiser.

Content farms and clickbait a love story.

Until now these fake websites were created through content farms. In them, workers with miserable salaries (you can imagine in which countries they are usually located) produce low quality content, most of the time copying content from other sites and putting "attractive" titles, you know. clickbait pure and simple.

The goal is to attract visitors to those sites, with minimal effort. That means copying content from other websites and altering headlines to attract attention. I'm sure you've seen headlines surfing the Internet such as "What's the point?this or that celebrity has died", has been arrested, or anything similar, being totally false.

11.9 billion euros thrown away.

Maybe at this point you think that the cost of having people putting content on junk sites may exceed the revenue that the "promoter" of the site could get, but according to Krzysztof Franaszek, founder of Adalyticsa digital forensic analysis and ad verification company, "The cost of content generation is probably less than 5% of the total cost of running a website.".

And this hole that ads are falling into doesn't appear to be a small one. The U.S. Association of National Advertisers discovered, through a study, that the 21% of advertising impressions were going to these types of websites. This could represent a waste of the advertising investment of more than 11.9 billion euros.

Now comes the horror movie.

If up to now we have been talking about web sites intended for hosting advertisements with low quality content introduced by people, now replaces "people" by AI.

According to a report of the media research organization NewsGuardThese websites are being populated with AI-generated text.

The result? A 140 major brands have been paying (and presumably, still are) for ads that run on AI-created sites, unknowingly, of course. And it seems that the 90% of major brand advertisements present on these AI-generated fake news sites were through Google.

An AI can create 1,200 articles per day.

To give you an idea of the scale of this activity, NewsGuard found that a single website was producing more than 1,200 items per day and that these webs were pushing the envelope by creating images and biographies of fake authors also by AI. To top it off, NewsGuard found that 25 new AI-generated websites were being created every week, which meant that it found 217 fake websites in 13 languages different since he started tracking this. 

For its part, MIT Technology Review decided to examine the list of nearly 400 ads from the 140 major brands that NewsGuard had identified on these fraudulent sites. It found that the affected sectors included finance, retail, automotive, healthcare and e-commerce.

And an inherent danger

It is clear that the only motivation of the ready The only way to make money with the least amount of effort is to make money. But the creation of these fake websites created with AI and seeking to attract clicks through sensationalist headlines, can spread disinformation.

A future Internet full of artificially created and low quality content only designed to attract your attention.

We are all aware that a scandalous headline is going to attract much more attention than a calm headline, so it's clear that the guidelines dictated to the AI behind these websites are clearly go for the clickbait. It is foreseeable that this will create a huge amount of false content. And not even the intention of these contents is the propagation of disinformation, after social or political interests. The only goal is economic. Visit my website, created by an AI, so that the ads served to me by Google are published and so I get income with the minimum investment of time and money.

"The opaque nature of programmatic advertising has turned major brands into unwitting participants, unaware that their ad spend indirectly funds unreliable AI-generated sites."

Jack Brewster, editor of NewsGuard.

A glimmer of hope

Despite this bleak outlook, Franaszek is cautious and believes it is still too early to tell how AI-generated content will affect programmatic advertising.

For these sites to make money, someone has to visit them. It has been shown that content farms exist because their promoters make money from it, otherwise they would cease to exist, but as this involves a cost of only 5% of the maintenance of these sites, it is unclear whether generative AI will eventually be the natural replacement for these farms. Let's keep our fingers crossed.

Cover image by Firmbee.com