The industry will never be able to fully define fraud. Like the concepts of love or human thought, ad fraud is hard to define specifically and exists on a spectrum rather than being black or white. The early days of ad fraud featured obvious fraud: single browsers bombarding bad websites with thousands of impressions from a data center. It wasn’t tough to draw that line. But, as fraudsters have become more adept at making fraud look like human traffic, it’s become more complicated. Now, wherever we collectively draw the line on fraud, it will be too liberal for some and too conservative for others. In short, there is no “right” definition, which only makes this problem harder.
Fraud-prevention vendors inherently have conflicting interests. To be clear, I’m not saying any of the existing vendors operate with anything but the best intentions. This is actually a problem in many professions. Your dentist legitimately wants you to brush your teeth after every meal and floss once a day, not eat too much sugar nor drink too much soda. However, if everyone was a model dental citizen, a dentist wouldn’t have a very profitable business. Fraud vendors have businesses to run, and wiping out fraud altogether would put them out of business. To be fair, if every product designer was so amazing that no advertising was needed for any product, we’d all be out of business! Understand this dynamic when determining which partner to work with.
The fraud problem is worse than most let on. There are certain things you can’t “un-see.” As I dug deeper into fraud, I learned that the spectrum is vast, there is a large percentage of undesirable traffic getting by every fraud vendor’s algorithm, and not enough people are doing enough about it. In fact, the first article I saw that publicly acknowledges the problem to the level I’ve personally seen was written just recently. I’m very surprised that most major marketers haven't looked hard at their raw data, because there is easily an additional 10% of traffic that should be removed that is still running for most agencies and advertisers today.
Pre-bid products are probabilities, not guarantees. When you hear about a pre-bid fraud prevention product, it’s easy to think, “Wow! This is like a bullet-proof vest for ad fraud. Sign me up!” But after signing up, the fraud only drops marginally. In fact, the money saved from not buying that fraud is sometimes less than the money paid for the pre-bid product. The reason this happens is that the pre-bid product is not evaluating the impression so deeply in real time that it can make a full determination as to whether or not it’s fraud. Instead, it relies on shortcuts to establish a probability. Sometimes it guesses correctly, and sometimes it doesn’t.
Direct buys are not immune to fraud. The press has so often equated ad fraud with programmatic that many now assume programmatic is the only place fraud exists. This is far from the truth. In fact, the Integral Ad Science Q1, 2016 media quality report reports fraud rates of 2.4% for direct buys made in the U.S. While lower than the 8.3% overall rate, it’s likely many of you are seeing rates similar to 2.4% or lower from your programmatic partners simply because those partners have taken great strides to clean up the traffic they receive and bid on. Fraud in direct buys is most often through sourced traffic, a tactic employed by most major publishers.