Here’s the thing: ad fraud isn’t going away. In fact, it’s only expected to evolve. By 2023, global ad fraud will reach an estimated $100 billion, according to Juniper Research.
Before that number rises, it’s in advertisers’, agencies’, and publishers’ best interests to get protected now. Big brands and agencies most likely will team up with name-brand ad fraud solutions. But name-brand doesn’t always mean top-notch performance.
To keep things simple for advertisers, many big-name ad fraud solutions rely on only a handful of key metrics to judge whether or not traffic is fraudulent.
One common metric is viewability, a concept developed by leading advertising organizations IAB and MRC. In theory, viewability is supposed to measure ad impressions that are actually seen by users. Desktop display ads are considered viewable if they meet the following criteria:
Standard Formats. 50% of an ad’s pixels are in view for at least one second.
Large Formats. 30% of an ad’s pixels are in view for at least one second.
Video Ads. 50% of an ad’s pixels are in view for at least two seconds.
However, the viewability metric doesn’t determine who or what made the impressions, be it humans or bots, real or fake. No matter what, advertisers still have to pay for those “viewable” impressions.
It really shouldn’t come as a surprise then that fraudsters have ways to work around it. If they have the technical skill, it’s easy for fraudsters to program bots that can hang out on a page or watch a video for two seconds, triggering a payable event. We’ve already seen this happen with the 2016 Methbot fiasco.
Probability and Percentages
Some ad fraud solutions use probability statistics to estimate the likelihood of a user being fraudulent. But probability isn’t definitive. What should an advertiser do if the solution says a user is 50% likely to be bad? After all, there’s also a 50% chance the user’s real, right?
Lots of advertisers are totally fine with taking a risk, but others don’t want to play that game, especially if they’re trying to stay TCPA compliant. As a precaution, advertisers often become hyperfocused on clean, premium traffic, thinking that they’ll lessen their chances of getting bad leads or bots.
Some solutions offer traffic scoring as part of their packages. Traffic quality is measured using a pyramid model: a small amount of highly-scored traffic sits at the very top, while low-scoring traffic overwhelmingly rounds out the bottom.
Naturally, most people want the good stuff, but there’s only so much high-quality traffic available. The laws of supply and demand take effect, driving up prices for premium traffic. And a higher price doesn’t guarantee conversions, either. Meanwhile, bad traffic only gets cheaper.
Performance Is Key
There’s no perfect ad fraud solution out there, unfortunately. But there are a few things advertisers can do to find a solution that fits their needs. Having one tried-and-true solution in place is a good start, but you should consider other options as time goes on. Ad fraud is fickle, and not every solution can keep up with the changes.
When testing different solutions, measure traffic performance by keeping track of something more verifiable than impressions: conversions. If the solution is working correctly, the traffic it lets through should perform well or at the level of previous benchmarks. You’d notice a change in conversion rates if bad traffic snuck through or good traffic was filtered out.
Ad fraud software is clearly necessary, but advertisers need to know how these tools work and not just pick one because they’ve heard of it before. Only then can they adopt better strategies and use ad fraud solutions they way they were intended – to help keep budgets in line and ad campaigns safe.