AppsFlyer’s recent #FoolsNoMore initiative has pushed a number of facts about fraud into the limelight. There’s a number of facts and trends amongst mobile fraud featured in the report that showcase the ever-changing battle lines of fraudsters and the tactics used to fight them, but chief amongst them is the rise of the bots.
As part of our ongoing conversation on fraud, we thought we’d dig a little deeper into AppsFlyer’s report and examine what makes bots so appealing to malicious actors, as well as what’s being done to stop them.
Rise of the Machines
The increased automation in mobile advertising is a double-edged sword.
On the one hand, AI optimization of ad creative, targeting, and lightning-quick programmatic exchanges are making mobile ads and digital advertising, in general, more cost efficient for advertisers while delivering peak revenues to publishers.
On the other hand, the automation of clicks and fraudulent actors has let to a massive spike in fraudulent activity. According to AppsFlyer’s State of Mobile App Fraud report, bots surged to replace click-farms last year — over 30% of fraudulent app installs come from bots as of February 2018.
From some enterprising reporting from Bleeping Computer, anyone can rent a network of 200,000 bots for a minimum of two weeks for any reason (inflate their own revenue, drive their ad price in a specific direction, or even more nefarious needs).
Luckily for advertisers, attribution partners like AppsFlyer are able to identify behavior patterns that differentiate real users from bots. As bots and scripts have advanced, so too has their ability to simulate normal user flows and behaviors but the fraudsters’ desire to optimize still makes bot activity stand out after enough analysis.
Greed makes them stand out.
Things like time to install, time to launch, and other measurements of how a user should behave give everyone in the mobile advertising chain information they can use to identify, quarantine, and block malicious “users.”
Furthermore, many ad networks and attribution partners go through serious and rigorous certification processes like TAG Certified Against Fraud, and direct integrations and seals of approval such as MOAT, Integral Ad Science, and DoubleVerify. These third-party verifications assure advertisers that they’re doing everything in their power to identify, track, and eliminate as much fraud as feasibly possible without tipping their hand to fraudsters.
Besting the fraud bots, like any other solution to a complex problem, requires constant analysis, testing, learning and relearning from experiences.
Simple Scripts Making Big Problems
So what are bots, and why are they the new preferred method of app install fraud? They must be really complex, IBM Watson level supercomputers to create such a problem.
It’s easy to see just what a big problem mobile fraud is, and the lengths ad networks like AdColony, and attribution partners like AppsFlyer are doing to fight it. There’s even the Coalition Against Ad Fraud, led by Adjust; a cooperative group of ad networks and other mobile ad industry companies that would otherwise be competitors working together to fight this massive problem.
The thing is, those bots aren’t all that complex at their core. It’s not about quality, it’s about sheer quantity.
They’re not really robots or even complicated pieces of AI. In most cases, they’re simply automated processes running over and over until someone tells them to stop.
While bots can be based on real phones, most bots are server-based, which means fraudsters can spin up more bots than ever before, testing new methods of attack without the hard limit of requiring physical devices. Bots are designed/scripted to send clicks, installs events, and in-app events for installs that never really occurred at all.
You can’t install an app onto a device if the device isn’t real, can you?
Join the Conversation
What’s your best anti-fraud fact from the AppsFlyer report? Tweet us at @AdColony. For the latest AdColony mobile news and updates, follow @AdColony on Twitter, like us on Facebook, or connect on Linkedin.