Misconceptions About Mobile Ad Fraud
In principle, user acquisition (UA) managers have a simple aim: gain more money from the new users, than the money spent to acquire them. In other words, have a positive “return on ad spend” (ROAS). The practice is, of course, much more complex with different sources, payout models, measurements optimizations, etc. And then, there is fraud, which messes with everything by wasting budgets, decreasing ROAS, polluting measurement data such as retention and LTV calculation, etc.
There are some precautions that UA managers can take of course, but there are also misconceptions that may, in fact, expose them to more fraud.
Let’s talk about some of them to have a clearer picture.
Misconception #1: “Click volume is not important on CPI campaigns”
In its essence, running a CPI campaign is not for worrying about the number of clicks, because only installs are being paid. However, a high or low number of clicks can be an indicator of fraudulent activity.
Device farms, bots, and unintentionally incentivized traffic generate a high click-to-install conversion rate; which means more clicks can convert into installs. So, the low number of clicks compared to acquired installs is an indicator of these types of fraud. The high number of clicks can also be a way to steal attribution from other channels, like click spamming which is explained in the next misconception.
Realizing the above, fraudsters can hide their footprints by sending a high number of clicks before the installs, or they can generate fake installs after a high number of clicks. Thus, it is better to analyze the click and install patterns along with numbers to detect fraudulent activity.
Misconception #2: “CPA buying is safe”
At first glance, buying in-app events instead of installs looks like a safer choice; because it transfers the optimization load to the ad network. It also seems like it transfers the fraud risk to network partners as well; because you are only paying for in-app events such as “purchase”, right? Unfortunately, that is far from true.
There are different ways fraudsters employ to steal your organic installs and they are called “attribution frauds”. Their methods aim to fool attribution tools to thinking that they provided the last click before install. It can be done by sending unintentional clicks from real user devices (click spamming) or perfectly timed clicks right before the install (click injection). These methods force advertisers to assume that they acquired a highly engaged user from a paid channel, while they acquired an organic user which would have been acquired anyway. Since these users tend to show high in-app engagement, they generally provide the targeted CPA as well.
If a CPA campaign gets subjected to an attribution fraud, which has a higher probability than CPI campaigns, in-app events that are paid for can actually be wasted budgets on organic users.
Misconception #3: “High in-app performance means minimal fraud”
Similarly, high in-app performance does not mean that there is no fraudulent activity mostly because of attribution frauds explained above. What is more worrying is the possibility that highly engaged users may not even be real users.
Fraudsters are employing more and more sophisticated methods in time. For example; they can now set their bots or simulators not only to install the app but also mimic humans to show better in-app activity. This causes advertisers to believe that they have found a high-quality source and they can allocate more budget for that channel.
Even worse, fraudsters can fool attribution tools to believe that an install or event has taken place by reverse-engineering their SDK-to-server communication and sending false signals. This method is called “SDK Spoofing” and it can mimic clicks, installs, events. Depending on the spoofer’s ability, it can get very hard to detect and a third party fraud solution should be used to detect anomalies.
Misconception #4: “Having transparency to see publisher app names prevents the fraud”
It certainly helps regarding campaign optimization and brand safety, however, it cannot be a reliable standalone datapoint in terms of fraud prevention.
In a better scenario, the names that are received as publisher apps can be used to create a fraudulent app name database. In that case, UA managers indeed have a chance to identify the apps that are sending fraudulent traffic and blacklist them from their campaigns.
In a worse scenario, those may not even be the real app names that the ad is shown on. It is just another parameter that ad networks pass on clicks and they can actually change and manipulate while redirecting the click. Unless you are sure that those are the apps that display the ad, it is not reliable data.
Misconception #5: “Fraudsters do not target low-volume advertisers”
It is a misconception because fraud has nothing to do with volume; it is all about profit. If the payouts are high, the campaigns can still be highly profitable for fraudsters. Even with low payouts and low volume, fraudsters are getting more advanced and fraud methods, even SDK Spoofing, is getting easier to conduct. Assuming low-volume advertisers are less likely to use a sophisticated fraud prevention tool, they are actually more prone to fraud than high-volume advertisers.
Misconception #6: “Monthly deductions are protecting marketing budgets”
Deductions and refunds certainly help the advertiser side to protect their budgets, but they can harm the advertiser’s budget in the long term, and they definitely harm the players in between the advertiser and the fraudsters.
Not all ad networks are conducting the fraud themselves and they are actually wanting to help advertisers to receive the highest quality traffic. If an advertiser does not provide feedback about the traffic they receive and they ask for a refund at the end of the month, this does not give ad networks to improve their traffic quality within the month. If deduction happens after ad networks make their payments, it actually causes them to lose money and fraudsters are still making money from it.
On the other hand, there are fraudulent ad networks that help or initiates the fraud. Not paying for the fraudulent install is sensible action. However, taking this action monthly can actually harm the legitimate partners, because they are the ones whose installs are stolen. It increases the CPI & CPA costs in the long run, harming the advertiser itself. You may want to read our article about this issue: https://interceptd.com/fraud-will-cause-your-cpi-costs-to-increase-soon/
Aside from the monetary loss, receiving fraud for a month pollutes marketing data such as retention and LTV, and makes it harder for advertisers to set business strategies.
The best way to prevent all these is to use a fraud prevention solution that works real-time and takes action real-time such as real-time install rejection and real-time sub-publisher blacklisting.
Misconception #7: “Brand safety and viewability tools are protecting against fraud”
Brand safety tools help advertisers to control where the ad is served. It is more common to use these tools on the web, whereas mobile is getting traction as well. Basically, it helps advertisers to ensure the ad placement is in line with their brand image. However, it does not ensure that placement is fraud-free. Many fraudsters, especially those targeting to steal organic attribution, are actually legitimate-looking apps that real people are using, and the app can be in line with advertiser’s brand image.
Viewability tools serve a different purpose; it ensures that the ad is visible by the user on the screen. However, this does not prevent fraudulent traffic as there is no control who sees the ad. For example; device farms employ many devices and they click on real advertisements to download the app in high volumes.
“Misconceptions of mobile ad fraud are limited to the cases stated in this article”
Unfortunately, no. But we will tackle this invisible –yet detectable– danger together. Next article will focus on other aspects of fraudulent activities in the realm of mobile ads.