Advertisers Can't Stop Fraud, But They Can Stop Making Costly Assumptions

July 11, 2017

By Rich Kahn, CEO and co-founder of eZanga



Advertisers are wrong about fraud: It's a problem that can't be completely solved. We are not going to eradicate it, the same way we are not going to wipe out robbery or eliminate illness. A more realistic — and effective — approach is to manage fraud, both by taking measures to prevent it, and by working to detect it once it occurs so that we can mitigate its most detrimental effects.

A new report from Digiday found that on average, 15 percent of ad network traffic is fraudulent. Putting an exact price tag on how much that costs advertisers is challenging, but irrefutably, it is a multibillion dollar problem.

By 2025, the global cost of ad fraud is projected to reach $50 billion. That is not to say our efforts are for naught. Research from the ANA and White Ops predicts that in 2017, bot fraud in advertising will actually decrease by 10 percent compared to 2016, from $7.2 billion to $6.5 billion. News like this is encouraging, but we have to be mindful of how we frame it. Bot fraud may be in decline, but what about human fraud, in which real people generate fraudulent traffic? This is often harder to detect than bot traffic.

We are not "winning the war;" we are getting better at fighting it. Anti-fraud technology, advertiser education and cross-industry, collaborative initiatives, like the Trustworthy Accountability Group, are all powerful and important, but they are not a cure-all. We, as an industry, have to be realistic. Doing so will help us combat fraud more effectively.


Dangerous Misconceptions about Fraud
The most widely-used tactics for fighting fraud are built on probability statistics, which means data is used to make calculated guesses, not definitive conclusions, about traffic quality. Let's say a solution indicates that there is a 70 percent chance of a traffic source being fraudulent. This information can certainly be useful, but advertisers need to recognize it as the solution's "best guess," not an absolute truth. In fact, a different traffic scoring system may have another interpretation about the quality of that traffic. This is not to say that these measures don't have merit, it's just that advertisers have to be mindful of their shortcomings.

Consider fraud prevention blacklists. While it's tempting to think about a computer as being either infected or clean, the truth is a device's status can ebb and flow. Suppose children jump on their home computer and download a ton of free games. Uh-oh. Now the home computer is infected with malware. Over the next few days, the parents notice it is running slower, but they move forward with their online purchases. However, when the problem persists a week later, they decide to run an anti-malware program and restore the system's good health.

Fraud is dynamic, and therein lies the shortcoming in the IP blacklist concept. An IP address encompasses all of the devices behind that single IP address, which often includes multiple devices across several users. Most of these lists are not routinely updated, so a home computer that was infected at the time the list was pulled but has since been cleaned is now permanently viewed as off-limits. That means advertisers could be missing out on potentially high-value traffic, throwing out the good with the bad and ultimately hurting ROI. The opposite scenario is true, too. A computer that looked crystal-clear two weeks ago could now be ridden with malware. We can't just build a list of "bad" IP addresses and think that's enough to outsmart the fraudsters.

Another way that advertisers and their partners fight fraud is by monitoring their traffic for suspicious patterns, which could indicate that they are dealing with a bot, not a human. For example, if clicks consistently happen every two seconds or conversions are occurring too quickly, you are probably dealing with a bot farm. This solution doesn't account for the strides fraudsters have made, nor the fact that they deploy a combination of technology and human touch to do their evil bidding.

There are click farms that employ actual humans to fill out forms, make calls and perform other actions we typically assume bots can't do. Another misconception is that these human click farms are always based in some far-off place like Russia or India. We tend to think of fraud as an overseas problem, but recent revelations about foreign hackers and fake news have reinforced that this is a costly assumption. Fraud can come from anywhere, and fraudsters are adept at duping U.S. entities into unwittingly delivering their fraudulent traffic. That means blocking traffic from certain high-risk areas isn't enough to curb fraud.

Digital fraud is a low-risk/high-reward endeavor, as far as crime goes. As hard as we are working to stay a step ahead, fraudsters are working just as hard to maintain their lead. Soon, technology will make it possible to convincingly simulate anyone's voice — a dangerous development at a time when we are increasingly embracing voice command technology for our devices.

The point isn't to be alarmist, but rather to understand that with innovation comes new fraud challenges — challenges we have to be prepared to manage. Fraud protection isn't a "set it and forget it" type of issue.Fraud protection isn't a "set it and forget it" type of issue. It is fluid. We aren't going to wipe out this billion-dollar problem overnight. By realizing this, marketers can take a more realistic and effective approach to curbing fraud, and to protecting their advertising budget.


Rich Kahn is the CEO and co-founder of eZanga, an online marketing firm. Follow him on Twitter at @RichKahn.


The views and opinions expressed in Marketing Maestros are solely those of the contributor and do not necessarily reflect the official position of the ANA or imply endorsement from the ANA.

You must be logged in to submit a comment.