Founder & CEO at Kochava, a leading real-time data solutions company for omnichannel attribution and measurement and data marketplace.
getty
Given all the changes digital advertising has undergone in the past year with Apple and Google announcing major platform changes, staying vigilant against ad fraud may have taken a back seat. Unfortunately, however, ad fraud is never going away. Digital advertising operates within a complex system that has many loopholes where fraud can infiltrate, but with some common sense and guidance, advertisers can mitigate fraud that makes it past prevention measures.
The presence of mobile ad fraud has often been attributed to the last-click attribution model, and while influencer reporting and multi-touch attribution (MTA) are available, the last-click model, with one single winner, has remained. The problem that some of us in the industry have pointed out is that last-click attribution is a flawed, incentivized system susceptible to fraud.
It’s not just fraudsters who are to blame for fraud but chief marketing officers (CMOs) fixated only on “conversions” or resulting app install numbers, ad networks looking to falsely claim outcomes from insertion order (IO) demands from advertisers, and adtech providers that charge on a cost-per-install (CPI) basis. I’m not saying that these groups are fraudulent, but some of their focus and strategies may inadvertently invite fraudulent tactics and deliver misaligned incentives.
Every year a new estimate projects billions to be lost. Yet some consider the budget wasted on ad fraud as part of “the price to pay” when advertising at scale. Others forgo protective measures for lack of resources until an outrageous invoice forces them to investigate.
My goal here is to shed some light again on saving ad spend that doesn’t take much time or resources from your marketing team. There are great fraud prevention tools on the market (of course, I’m biased), but these tools search for specific behaviors and patterns that fraudsters work to outrun. Fraud is a moving target, and it often hides behind the performance numbers CMOs want to see in the first place.
Some of this advice might hurt, but it will save you in the long run (either in ad spend or reputation). The goal is to not wait until something catastrophic happens to force you to pay attention.
Have An Anti-Fraud Strategy
To start, if you don’t have a stated strategy — and the associated tools to implement that strategy — I’d advise starting there. If you don’t have tools already established, run through a request for proposal (RFP) with a select number of vendors and ask the vendors what questions you should be asking the other vendors in the evaluation. Part of the strategy should clearly separate viewability fraud (typically used when paying on a cost-per-mille [CPM] basis for your media) and attribution fraud (typically used when paying for media on a cost-per-click [CPC] basis).
The stated strategy could organically come from the RFP process, or you may have it mandated by executive leadership. The point is, use the strategy to back into the measurement framework and leverage the ecosystem to help you ask the right questions around both strategy and tools. Although there is no foolproof way to prevent media fraud 100%, implementing these measures will help you become more adept at identifying fraud.
Perform Quarterly Stress Tests
What I mean by this is to simply take a cohort and see what kind of ad signal is being received in aggregate. In mobile, the industry values the real-time feedback loop, but evaluating a cohort at the macro level is when one might see fraud creeping in and show some opportunities for improvement. For example, certain data patterns could point to misattribution as a result of install sniping. This is especially evident if you see much activity past the initial campaign launch because the farther out you see installs, the more likely that they are organic.
Look Beyond Install Numbers
Dig into your conversion and post-install event rates by cohort. Here is when you are more likely to see anomalies such as a device ID or email that looks off and from there investigate behaviors associated with these potentially false identifiers. Other metrics to evaluate include time to install (TTI) and time to engagement (TTE). Installs and completed conversion events occurring too fast or grouped can be flagged for further evaluation by an analyst. These are some of the many metrics we collect and correlate across the nexus of media supply when evaluating media fraud in real time.
Be Conscientious Of Your Audience Reach
There has been a large focus on user acquisition, but the farther your reach, the less reliable the traffic. Cheaper CPIs or a broad reach will make you vulnerable to fraud. There are only so many ad partners out there, but to fulfill demand and feed a client’s growth may lead them to procure more marginal traffic. Make sure you have traffic verification measures in place to block suspicious traffic.
Implement New Preventive Measures One At A Time
Often our analytics team has made recommendations to brands but may not hear that any changes have been made. Perhaps the client doesn’t make changes out of fear that those install numbers will drop and raise eyebrows about past performance. My advice here is to implement changes one at a time.
For example, run a backtest omitting a suspicious partner and see what your results are, or implement a new fraud prevention tool on a new or updated app. Chances are, the number change won’t be as drastic as you think. If your vendor partner can’t run backtests, find a vendor that can. This is an important capability when qualifying the proof of the toolset.
Realize The Power Of Mobile And Implement The Tools To Support It
Success isn’t automated, and your marketing strategy will always require revision. Don’t submit to fraud as an acceptable part of advertising. Instead, focus periodically on substantiating your campaign performance. Pull the curtain away and see what’s really going on behind the scenes. The truth might sting, but it matters.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
