By now, you can reasonably expect that marketers have heard about ad fraud and may even understand how damaging it can be — from wasted ad dollars to the systematic funding of fake news and hate sites. But yet marketers keep buying large quantities of digital ads through programmatic channels. I have heard every excuse they use to justify their continued purchases, despite the fraud. Let’s look at a handful of these more closely, to see the difference between what they assumed to be true and what is actually happening.
But we get more reach…
Most big advertisers turned to programmatic buying because they couldn’t get super large quantities of ad impressions to buy from big, mainstream publishers. Ad exchanges promised them vast quantities of ads, on vast quantities of sites, targeting vast audiences. Hmmm. Something should have already smelled fishy, right there. But marketers assumed they were getting more “reach.”
But the question is whether that reach included any humans! If they understood how ad fraud works, marketers would know, it is easy for fraudsters to set up large numbers of fake sites, use large amounts of bot traffic, to generate vast amounts of ad impressions to sell. If advertisers bought these impressions, none of the ads would be shown to humans. The reach is not human.
Marketers then say, “oh, it doesn’t have to be all humans because the prices are far lower; so fraud is priced in and I’m ok with that.” The prices of programmatic ads are indeed lower than buying directly from good publishers. Good publishers have real costs to create content that humans want to read. They have finite human audiences that visit their sites a finite number of times. So the number of ad impressions they can sell is finite too. Simple supply and demand would tell you that more limited supply means higher prices.
Fake sites, however, have no cost of content. They either plagiarize all the content using web scrapers (software programs that load others’ web pages to copy the content), or they create all the content from scratch using algorithms. Because these fake sites have no costs, they can sell their ads for low prices and still be highly profitable. They deliberately under-price mainstream publishers so advertisers would allocate more budget to them, chasing the lower prices. But marketers get what they pay for. Low prices = low quality or no quality (i.e. no humans at all).
But we get higher engagement…
Marketers then say they buy programmatically because they get higher “engagement.” This word is in quotes because it is not real engagement — i.e. humans liking and clicking the ad. The clicks are done by bots to make it look realistic and desirable. If there were zero clicks for months, marketers would pull budget away. So bots click. Then bots click more to trick the buyers into allocating more dollars to them because they think there is higher engagement. Yay, bots engaged with your campaign! (But humans did not).
But fraud detection said there was no fraud…
Backed into this corner, marketers then start to shift the blame elsewhere. They say the fraud detection companies tell them fraud was low in their campaigns. Trade associations told them fraud was low too, like 1.53%. And exchanges even offered them fraud-free guarantees so they are assured of zero fraud! Basically, someone else told them not to worry, and just keep buying.
But what if the detection technology couldn’t detect the fraud. What if the bots were advanced enough to trick the detection and get marked as “valid.” Or what if the bot, or mobile app, simply blocked the detection tags, so no measurements could even be made? Is that low fraud, or just undetected fraud? Trade associations and ad exchanges cite and use these same verification vendors. When they offer fraud free guarantees, that’s only as good as what the tech can see. Sure, they are happy to give you back 1.53%, if you find the fraud. But you can’t verify they got the number right and you didn’t find the vastly larger amounts of fraud that were invisible to their tech.
But blockchain…
