When digital publishers hear the words “ad fraud” their hearts tend to speed up a little bit. This isn’t because it is some scary thing (well, maybe it is) but it is mainly because when it comes to ad fraud, there is very little people can do to fight it.
What is Ad Fraud?
Ad fraud, for all intents and purposes, is when both human and nonhumans create traffic for the sole reason of being misleading. This can be done in a variety of ways, and the worst part is that technically it’s legal.
Because there is no risk of going to jail, the amount of money attributed to fraud is on the rise. With Statista estimating that by 2022 $44 billion will come from ad fraud. That’s a 214% increase since 2017!
While some people may see one very evident benefit to this kind of fraud (hello pay day!), there are some big down sides to it as well. Marketers specifically lose out on fake conversions and bad brand placement. Users experience an excess of ads, click bait journalism, fake news, and malware. Advertisers start to lose the trust of their partners as fraud rates increase.
Types of Ad Fraud
Nonhuman Traffic (aka bots) is most likely the first this that pops into a person’s head when they think about ad fraud. Bot traffic, is meant to scour the web generating fake impressions and clicks. Bots can come from the simplest static IP that are easy to track and block, to a more sophisticated rotating IP bot that mimics real movements.
At Mobile World Congress back in February of this year, a panel of experts talking about artificial intelligence said that they don’t even believe AI can stop ad bots. Strategy Lead at White Ops, Jay Benach, mentioned how the US government took down a large bot that had infected 1.7 million machines in order to mimic human behavior. And that was even after the firm introduced a technology they thought had a handle on the problem.
Human Traffic is much more complex and comes in a much larger variety of forms. In general, human traffic revolves around two characteristics: 0% viewability, or intentional misrepresentation. For example, invisible ads is when one ad is hidden or “stacked” behind another to generate a false impression. There are also more serious cases, like arbitrage, where a group purchases traffic at an extremely low cost and then resells it for double (if not triple or higher) the price.
Eating Up the Market
In a webinar organized by Adjust and Liftoff, titled Mitigating Mobile Ad Fraud, Andreas Naumann outlines the 4 major areas we are seeing apps identify fraud.
SDK Spoofing accounted for 37% of fraud in 2019. This works by bad actors spoofing installs based off of previous interactions in order to trick the SDK. Naumann then goes on to show how they use encryption on every request they receive in order to counteract fraud. The model they use is similar to other networks on the market.
Click Injection came in second with 27% of fraud in 2019. While devices like android are trying to counteract this in newer updates, older models are still at risk. Click injection works by inserting an ad that isn’t in the publishers inventory without their knowledge. Key indications that this may be happening are if you see close to 100% conversion rates.
Fake Installs is a close third with 20% of fraud in 2019. Fake installs occur when fraudsters emulate users and install apps on fake devices.
Click Spam is last with 16% of fraud in 2019. Click spam is one of the oldest forms of ad fraud, as it has been around since the 1970s. It goes hand in hand with schemes like click farms, that hire people for the sole purpose for clicking on ads and filling out forms. This devalues a publishers impressions, clicks, and conversions.
Of course, these rates only come from the percentage of fraud that is actually detected by ad networks. On the programmatic side of things, ad fraud is scaring away marketers. However, it doesn’t just happen in programmatic buying, it can occur in direct buys as well.
10% of direct buy desktop displays and 6% of direct buy mobile displays were attributed to ad fraud in 2017. Most of this comes from what Naumann calls “noncompliance fraud”, where a partner breaches contract agreement.
Well, there may not be an end to ad fraud in sight, but the prevention of fraud is getting more and more sophisticated.
In the same webinar, Matt Fisher of Liftoff states that the best ways to counteract fraud is to monitor at the top of your funnel. If any request has missing information, it may be cause to not fill it. He also mentions that you should ignore bid requests from publishers who are requesting too many ads per day. These are clear signs of fraudulent behavior.
Further down the funnel, Fisher says that you should look for any metrics that are “too good to be true”. Extremes are always bad, like click through rates lower than 0.1% or higher than 15%.
When it comes to choosing a network to partner with, you should make it best practice to ask as many questions as possible. Discuss what their ad fraud policy is, their pricing, be sure to define what you are looking for in a partner (not just what you don’t want), and above all make sure you have complete transparency.
Supply Path Optimization
One way both SSPs and DSPs are implementing to fight ad fraud is supply path optimization (SPO).
What is that? Well, Digiday defines it as a solution that “helps improve bidding efficiency in online auctions and also highlights any shady supply-side platform tactics”. As the industry is pushing for transparency in the digital marketplace, a solution like this is extremely important.
Take AppNexus’ model: SPO looks at a publisher’s SSPs and analyzes them for traffic and win patterns. By doing this, they can quickly spot any partner that is using aggressive tactics and shut them off; allowing for a safer, more streamlined process.
Because of this, buyers are comfortably able to bid and win auctions at reasonable prices and publishers are able to maximize revenue. Win win!
No matter how we fight it, ad fraud isn’t going anywhere any time soon. The best we can do is try to partner with companies who implement ways to detect it and continuously monitor our metrics to keep our brands safe.