AppLift recently announced the release of its advanced fraud fighting suite – the Fraud Buster. The topic of ad fraud has been covered quite excessively in mobile industry-related news recently, as it is now more pressing than ever.
Fraud has been around since mobile advertising picked up and from the looks of it, it’s not going anywhere. We’ve seen different manifestations of mobile ad fraud and with fraud solutions becoming more advanced, so has fraud. It has undergone a tremendous evolution from the early stages of simple Bots and Auto Redirects. Fraudsters have since developed more sophisticated techniques such as click spamming, ad stacking and click injection. With fraud having become more clever, there is a need for fraud solutions to go a step further.
While there has been a greater focus around fraud prevention and detection, most typical approaches to fraud fighting remain a retrospective analysis of fraudulent patterns. This approach has a big weakness in that it can only ever analyze fraud after it has already happened and can only take reactive measures against it.
So, What is the Fraud Buster?
Fraud Buster has been developed to fight fraud in real-time at install level. It evaluates each install individually and applies a set of rules to determine whether to approve or reject those installs. For years we’ve been fighting fraud at AppLift with a combination of statistical modelling, data evaluation and human experience.
Over time, we realized that we needed to go a step further. As a demand-side company our efforts are focused on delivering value to our advertisers, and preventing fraud is at the top of that commitment. What our advertisers needed was a proactive approach for real time fraud fighting along side with pattern analysis and offline traffic controlling. This is what the Fraud Buster does.
How Does the Fraud Buster Work?
There is a set of rules that the Fraud Buster uses to determine the validity of an install, such as Click-to-install-time (CTIT) limits, CTIT distribution and pattern evaluation, IP addresses (same IPs, IP Blacklisting), as well as Referral keywords to filter out non-compliant traffic. An additional value add is that thresholds for each of these rules can be set individually by the advertiser. Based on the size of their app, and other criteria specific only to them, they can exercise utmost control over which installs to approve and reject.
As an example, an advertiser could set an individual CTIT threshold based on their app size. Let’s assume this threshold was set at 20 seconds due to the app being heavy in size (this threshold is flexible and can be adjusted to the advertiser’s app requirements), then any install that was below that threshold when it comes to the Click-to-install-time would be rejected automatically and in real-time. This way, no retrospective analysis is needed for advertisers to remain free of fraud.
The Fraud Buster algorithm makes decisions based on conversion information, as well as by looking into the history of conversions to ensure that the right decision is made. We are continuously improving our algorithm and adding more layers to our decision-making process to make it even better than it is right now.
Why Should Advertisers Use the Fraud Buster?
Advertisers that are looking to go beyond a retrospective approach to fraud fighting will benefit from the Fraud Buster. It gives them control over what to consider fraudulent and helps them remain fraud-free from the start, rather than having to go to the hassle of retrospective analysis and claiming money back at a later stage. The Fraud Buster’s proactive approach to fraud fighting can result in higher lifetime value of users and maximized ROAS. Its integration as part of DataLift 360, our unified platform with access to all mobile advertising channels, ensures maximum transparency on the overall marketing activity and ultimately, the quality of the marketing efforts.
Learn more about the Fraud Buster here.
Get to know more about ad fraud types and prevention methods in our latest free eBook available here.