Fraud Prevention: Options, Pros & Cons
Blog post by Dmitry Isakov, CEO FraudScore
“Recently we at FraudScore have noticed that a lot of clients are asking more and more questions about fraud prevention considering that to “prevent” means to stop ad fraud before a conversion or an install actually happens.
I wanted to devote this blog post to fraud prevention in order to explain the complexity of this issue, show when and how ad fraud prevention works, how it can operate and what advertisers can actually expect from it”.
WHAT IS FRAUD PREVENTION
Let’s start with a small explanation. In general “fraud prevention” means that an anti-fraud solution (either third party or inhouse) is able to check a click and/or a conversion the moment it happens, decide if it is fraudulent and block it. If the click proves to be fraudulent then there is no need to direct it to the next step. The advertiser doesn’t pay for these “prevented” conversions because the fraud is stopped from getting to the campaign statistics.
It’s obvious why companies are looking for fraud prevention tools — the scale of fraud is terrifying and marketers are searching for ways of protecting their budgets. In the first half of 2019, FraudScore stats showed that 35.03% of all processed traffic was fraudulent.
So how does fraud prevention work? There are several options for prevention and, to be honest, they all have caveats and limits that advertisers need to know and understand. I’ll try to explain the most common fraud prevention technologies and give details for each type of the solution.
FRAUD PREVENTION OPTIONS
Fraud prevention capabilities vary depending on the type and stage of marketing campaign where the fraud occurs and which is to be prevented. In this article we will take a closer look at CPC and CPI/CPA campaigns.
Option number one is a good fit for mobile app developers — integrate MMP’s SDK inside of the developer’s app. In this case the MMP gets access to the app statistics, detects suspicious activities and blocks those that are found to be fraudulent. If app developers give that access to MMP then “fraud installs get cleared out of the reports” because those installs simply don’t make it to the official dashboards.
Option number two — use an anti-fraud solution to analyse clicks. For example some marketing platforms provide this kind of click quality “targeting”. In this case — real clicks are directed to target URLs and the fraudulent ones are sent to fallback URLs.
Option number three — analyze and reject conversions using a third-party anti-fraud solution. In this case an anti-fraud solution takes a couple of minutes to analyse a conversion and to automatically reject those that are fraudulent. If your anti-fraud provider support this technology it can be used for prevention purposes too.
TRICKY PART OF FRAUD PREVENTION
Ad fraud prevention is a tricky issue, as it’s been already stated. It requires deep understanding of the matter from both advertisers and anti-fraud solutions. I’ll give some examples of the cons that advertisers might face when using fraud prevention:
First, the timeframe to make a decision “fraud” or “not fraud” is very limited. In this short period of time an anti-fraud solution is restricted in its options to perform a thorough analysis. So as a result a number of false negative decisions is rather significant.
Second — fraud prevention gives no context for the conversion. An anti-fraud solution has neither any data about upcoming campaign traffic, nor data about upcoming inapp user actions; therefore, there is no way to perform a comprehensive analysis. For instance, inapp aсtivity can certainly help in fraud detection by providing context for each conversion, but when it comes to fraud prevention the lack of sufficient time to gather and process contextual data can limit the analysis in isolating fraudsters.
There is a very limited amount of parameters that can be used for “fraud check-ups” before an actual conversion occurs. For instance, an IP address might be checked against a datacenter blacklist but this is when we come to a roadblock as there is a limit of data which we can analyze at this point.
Also, it’s worth mentioning that companies who drive CPA campaigns are more frequently requesting fraud prevention tools — they need to clear the fraud before the conversion happens and advertisers will see it. And usually such companies are the ones who face the complexity of fraud prevention. The most common fraud types in CPA campaigns are click injection and click spamming. In both of these fraud types fraudulent conversions are done on real devices which belong to real users. And the users are not aware of what is happening to their devices. So all the device parameters are real. In a fraud prevention scheme — if such “user phone” takes part in attribution fraud, a third-party anti-fraud might simply “accept” that user as a real one. The conversion will be marked as “non-fraudulent” and charged to the advertiser if we skip the post-install analysis.
FRAUDSCORE PREVENTION SOLUTIONS
FraudScore always supports an open dialogue with our customers about all aspects of the fraud fight. Our solution allows clients to prevent fraud with high precision; however, we always recommend that companies and their representatives learn more to increase their understanding of this particular technology since ad fraud prevention has its nuances.
FraudScore offers two options for fraud prevention — SmartReject and SafeClick tools.
SmartReject helps to detect fraudulent conversions and it can be used for prevention because the sources won’t be notified* about the conversion [till express-analysis is made] and there is enough time (5…60 minutes) for a more thorough drill down. FraudScore provides a detailed report for each rejected conversion, the statistics for the fraud types involved and rejection reasons.
Second option — is our new solution: SafeClick. We’ve had a lot of requests for a tool that will not just identify fraudulent clicks but will make a redirect to the target or fallback URLs based on the quality of the click. The real clicks are directed to the target URL, while the fraudulent ones go to the fallback URL. This solution is a result of our efforts towards real time fraud prevention and is based on research that our data scientists have conducted to provide a reliable and useful tool for the marketers.
It is a good idea to combine prevention and detection methodologies. This will help to cut the obvious fraudulent clicks and not let them go further. But later, at the conversion and post conversion phase, we can apply more precise methods to detect sophisticated types of fraud. FraudScore platform can also be used for detection purposes, helping its clients to get a comprehensive understanding of the quality of the CPI/CPA traffic they are buying.
IN CONCLUSION
Statistics prove that fraudsters are constantly on the verge of the most up-to-date development and coming up with new schemes and approaches. The upgrades that they roll out call for higher anti-fraud efficiency requirements. We are all working to combat it but there is always room for revision and updates to anti-fraud platforms. And “ad fraud prevention” is one of the new turns of fraud fight.
We at FraudScore are developing an independent anti-fraud platform and we want the key industry players to fully know and understand all of their options. We invite you to direct your questions about fraud prevention and detection to our managers and we’ll be ready to show and explain it all to you. Contact us via sales@fraudscore.mobi.
*depends on the marketing platform