Fake sites, fake traffic, fake ad impressions

Websites that don’t have a lot of natural traffic, and even websites that do have real human audiences, are tempted to buy traffic from shady sellers to drive up their own ad revenue.

The software bots used to generate fake traffic and impressions can easily trick fraud detection technologies and appear to be real. They can also be programmed to click on ads, view videos, stay on webpages longer to manipulate bounce rates.

All of these actions are used to manipulate analytics in order to cover up the fraud. Furthermore, various forms of fraud don’t even require large numbers of bots to hit webpages.

Or in mobile, bad apps like the fake flashlight, alarm clock, or photo editing apps can load thousands of ad impressions in the background every hour, whether or not the app is ever used.

Combine this with fake mobile devices (software that is used by mobile developers to test their apps) and cybercriminals have an ecosystem of tools and techniques that enable them to commit ad fraud on an incomprehensible scale.

For example, what if webpages themselves redirect to other webpages in infinite loops? Typical bot detection technologies would not catch it.

Fraud detection cannot detect it, but fraud is still there

So why do various reports say that “ad fraud is non-existent” or that it is low and going lower? It’s probably because the fraud detection technologies cannot see the fraud, not because the fraud is not there.

In fact, case after case of massive ad fraud are still being reported, despite the widespread use of fraud detection.

In countries where programmatic digital ads are still growing rapidly and mobile is the dominant channel, there is evidence of rampant fraud that continues unabated.

What can marketers do to combat and reduce ad fraud?

If marketers are not alarmed about ad fraud yet, they should be. But it’s not cause for panic. Instead, marketers can use common sense and their own analytics to see if ad fraud affects their digital marketing campaigns.

For example, humans visit websites during waking hours and sleep at night.

So if the hourly traffic to a website remains exactly the same every hour of every day, something is wrong — there should be more traffic during the day and less traffic in the overnight hours.

Or if 10 different referring websites have the exact same number of visits, or bounce rates, or clicks on the site, something is wrong.

Humans don’t move that way, but entire botnets can be programmed to visit websites in exactly the same way. Marketers must insist on getting access to analytics and detailed reports.

With such details, common sense can help pick out what is obviously fraudulent; these can be simply turned off so no more money is wasted.