App Fraud Around The Globe
From Korea to Kentucky, mobile app fraud is costing marketers around the world a boatload of cash.
For mobile app marketers launching campaigns in different countries, it’s important to know where in the world you’re most at risk, and which types of fraud you’re most likely to encounter in certain markets in order to implement the most effective safeguards.
In this post, we’ll give an overview of the most high-risk markets, as well as the different anti-fraud mechanisms at your disposal as you launch mobile campaigns in new markets.
First, let’s dive into common mechanisms used by third-party anti-fraud solutions to detect and prevent mobile app fraud:
IP (or Internet Protocol) addresses are temporary identifiers assigned to each device on a network. When a mobile app install is registered, marketers typically collect the IP address of the device that installs the app. In order to filter against IPs that don’t belong to real users or are mislabeled or masked, analytics providers maintain large, self-updating lists of IPs associated with data centers, TOR exit nodes, VPN services and other IP masking services. Analytics providers use proprietary heuristics to identify fraudulent IPs across their customer base and add fraudulent IPs to so-called “blacklists”. Any install associated with an IP on an IP blacklist is rejected, ideally keeping your campaigns free from fraudulent IPs.
Mean time to install or “MTTI” refers to the mean time between a user clicking an ad and installing an app. MTTI tends to vary by app vertical and network — for example, some apps tend to have a low MTTI because users download the app right when they need it.
Advertisers and fraud mitigation experts monitor MTTI closely as irregular MTTI distribution for a given app can be an indicator of click injection or click spamming fraud. For instance, in the case of click injection, fraudsters detect when apps are downloaded on a device and trigger clicks right before an install completes, thereby stealing credit for the install from the rightful source. In such instances, conversions may appear to occur within an extremely short MTTI, signaling suspicious activity.
Yet, as with most detection mechanisms, sophisticated fraudsters have caught on to MTTI outlier detection and have become increasingly adept at outsmarting this form of detection by obscuring MTTI anomalies. In response, app fraud mitigation experts must constantly upgrade their distribution modelling in order to catch this type of trickery.
The majority of installs happen within very close proximity to an attributed click. Thus, a common indicator of a fraudulent activity is when an install occurs far from the geographic location of the corresponding click. Some analytics providers perform statistical analysis on the geographic distance between installs and clicks for a given app, and when statistical anomalies arise, providers flag these installs for inaccurate geo-targeting (when a source fraudulently runs ads outside the geographic target specified by an advertiser) or as downright fake installs.
Too Many Touch Points
When an app receives an irregularly high number of clicks from a given source it can be a sign of click spamming. Click spamming is a fraud scheme designed to poach organic or paid installs. Bad actors send fraudulent clicks to an advertiser in the hope that by random chance one of the clicks is credited as the last click before an app install. To prevent against this form of fraud, advertisers and their analytics providers often measure the number of matching clicks for a given install and flag app installs where the number of clicks exceeds a certain threshold.
Now that we’ve established the most common fraud detection mechanisms, let’s dive into some markets around the world where advertisers are most at risk of app fraud.
Across the globe, mobile app fraud is distinctive from financial fraud in that app fraud is actually more prevalent in developed nations. Research shows that countries with the highest cost per action (CPA) and cost per install (CPI) payouts have higher fraud rates. After all, the incentives for fraudsters are the highest in these regions, while the consequences of getting caught remains the same.
Such high-reward regions include countries like Germany, Australia, the US, UK and China. Meanwhile, countries with relatively low payouts have a lower fraud rate — including countries like Indonesia, India, Brazil, Vietnam and Thailand.
China, in particular, has seen digital ad fraud surge in recent years. According to China-focused analytics solution AdMaster, in 2016, invalid traffic increased from 28.9% to 38.7% of total advertising traffic.
Another study, produced by fraud solution WhiteOps, compared ad fraud rates among European sources to fraud rates among U.S. sources. The study found fraudulent traffic from the U.S. is “significantly higher” than European nations. Specifically, 21 percent of impressions from U.S. domains appear to be fraudulent invalid traffic from non-human sources, concluding fraudsters have “infiltrated the U.S. advertising ecosystem to redirect large amounts of advertisers’ budget and traffic to bots.”
Traffic sources in Latin America fared relatively well, especially compared to North America, various studies have shown. One such study, conducted by antifraud solution DoubleVerify, found just one percent of display ads in Latin America are fraudulent, compared to the U.S., where six percent of display ads are fraudulent.
Notably, DoubleVerify found that across the globe, fraud arising from bots has started to decline as anti-bot solutions and ad-blocking software have started to take effect. Yet “infected browsers” are emerging as the new leading cause of fraud. Instead of bots creating “ghost” browsing sessions, fraudsters are now manipulating a user’s active browser, making it harder to pinpoint and prevent.
Wherever you’re targeting new users or user re-engagement via paid ads, app fraud will inevitably rear its ugly head. And while it’s somewhat counterintuitive, data shows that regions with the highest-spending users are actually the most susceptible to app fraud. Sophisticated marketers must keep these trends in mind while launching global campaigns, and equip their teams with the latest detection tools to prevent fraud from harming performance and adapt to fraudsters constantly-evolving attack patterns, wherever they might emerge.