The Eternal Struggle Against Fraudulent Advertising

Early this month, the Association of National Advertisers had some good news for people at Cannes Lions. A shocking 15% of all ad spend in 2023 went to media buys on mostly fake made-for-advertising (MFA) websites. That went down to 4% after a year. The news is a little less good because digital ad spending is expected to rise by 16% in 2024, which is only 4% of a bigger number. Also, 4% is still a lot.

Of course, one type of ad fraud is tricking brands into buying programmatic advertising on sites with little or no interesting content. Ad fraud is still a huge problem, even though different people have different ideas about how big it is.

Some types of ad scam, besides the use of MFA sites, are (but are not limited to)

• Click fraud is when bots are used to make a lot of clicks on pay-per-click (PPC) ads to make it look like they are reaching a lot of real people.
• Domain spoofing is when a fake high-value name is used. Even though clicks and users are real, they don’t mean you’ve reached a valuable group.
• Pixel stuffing is when several ads are squished into a single pixel frame. People who click on the frame are actually clicking on ads that they can’t see.
• Ad stacking is a lot like pixel stuffing, but the ads are the right size. The only one that can be seen is the one at the top, but clicking on that one clicks on all of them.

A big deal

If you think that ad fraud isn’t a big deal, look at some of the bigger figures of how bad it is. One group of researchers at Juniper Research said that this year, ad scams will cost about $100 billion, or 22% of all the money spent on media. The number is expected to stay about the same (23%), but since media spending is going up so quickly, that would mean a loss of $172 billion. That is still a very large number, even if Juniper is off by as much as 50% to estimate the loss.

“Half the money I spend on advertising is wasted,” to quote John Wanamaker. That should be 75%.

Ann Tarasewicz, CEO of programmatic platform Axis, gave us the Juniper report. When we talked to her, she was briefly living in Romania instead of her home city of Odessa in Ukraine. „The level of ad fraud we see now is a big problem for the whole business. The con artists are trying to get smarter, better planned, and more up to date.

How can brands put up with this? “Of course, brands care most about volume, but I would say that if a publisher is involved in low-quality traffic, that can hurt the brand’s reputation.”

Are people paying attention?

In January, Greg Krehbiel, a writer to MarTech, used another amazing number from Bob Hoffman’s book “Adscam”: Only 3% of the money spent on internet ads is actually spent on ads that people see. “No, that’s 97%.”

Krehbiel said, “It’s a really bad problem.” “Before I published that article in January, there was a story in Forbes about how they made a whole website with ads that no one saw.” Forbes had this page up for a few years “by mistake.”

Could it be that brands don’t mind wasting money on ads as long as strategies are working? “It doesn’t matter how much fraud is in it,” Krehbiel said. “If it’s 99% fraud, the 1% is good, so what’s the problem?” I believe it all comes down to whether people decide to keep running a campaign based on numbers from the ad providers or on the results they are seeing on their own.

What should be done?

Tarasewicz gives a number of ways to stop ad scams. “Sorry, but there isn’t a solution that works for everyone,” she said. “Brands should work together to solve this problem.” Her favorite way of working together is through groups like the Trust Accountability Group (TAG). “It brings marketing companies, book publishers, police departments, and everyone else in the market together.” By sharing what they know, they can figure out better ways to stop scams. It was started in 2015 and works pretty well.

Artificial intelligence, fraud detection tools, and ad verification partnerships are some of the specific ways that fraud can be stopped. This method uses machine learning, data analysis, and pattern recognition to check the validity of impressions, clicks, and conversions. The goal is to find and stop fake traffic in real time.

The AI method

Another issue with AI is that scammers can also use it. Tarasewicz said, “They are trying to be up-to-date here.” Fraudsters are trying to test and adapt while brands are making changes to their systems. But even if they can get used to one system, a different one that tracks scams might give them a different answer. Because of this, it is best to use more than one method.

Bots that are more advanced can now do more than just click and make marks. They can now also place orders or fill out forms. “This is really hard to figure out whether it’s a person or a machine.”

When banks use AI to find financial scams, it seems to work well. Krehbiel said, “That’s different because all the transactions are coming together in that case.” That’s the problem with ad fraud: they say there was a transaction, but I have no idea if it happened.

The method for verifying ads

It can be helpful to work with ad verification companies and add their services to ad exchange sites when it works. Recently, even a verification partner with a good reputation had to say sorry for mistakes. Tarasewicz said, “The different ad verification companies use different methods, and we don’t know what they’re using.” That’s what I think you should do: use more than one tool, at least two. It’s a shame that various tools often yield various outcomes, especially when it comes to various types of ads and visits.

Tarsasewicz knows of an ad verification service that her customers use. It doesn’t work for scanning CTV traffic but is great at checking in-app ads. Other tools, on the other hand, are reading CTV correctly, but the results aren’t shown for 60 days. In the market, there are different cases. It is suggested that you look at how well-known the tools are on all of these platforms.

Krehbiel said, “I don’t want to be too negative, but the idea of an ad verification network trying to figure out if an ad is being shown to humans is flawed because the network itself is a bot.” It’s definitely better than nothing.

Tarasewicz and Krehbiel both agreed that these tools probably won’t work well in all situations. “No one can know everything there is to know about ads on Twitter, Facebook, and the web.” It will depend on whether they’re looking at B2C or B2B, as well as whether they’re looking at broad or specific customer interests.

There are more than one tool in the kit.

Would Krehbiel still be negative about the situation if he followed Tarasewicz’s advice to use a mix of prevention methods? He replied, “Yes and no.” “That tells you something if several different signals all point in about the same direction.” It would be rude for me to disagree with an ad network if they said they were showing my ad so many times and another ad fraud tracking group said the same thing.

Of course, it brings up the cost issue. “You’re not only making the ads; you’re also looking for ad fraud.” When you figure out the ROI for the ads, that changes. Look at this Super Bowl ad for Pepsi. How they judged success was by how much Pepsi they sold. I think that is a much better way to figure out if an ad is doing something than these tech options.

Just so you know, ads do work. Seeing an ad and then buying something is something that everyone has done. You’re not just yelling into the void. Krehbiel said, “We know for sure that advertising works and can work.” “The real question is: Does it work as much as they say it does?”