Cost Per Install (CPI) fraud is a growing headache for mobile advertisers looking to generate installs and expand their mobile user base. Bots, cybercriminals, and other abusers are taking advantage of incentivized marketing campaigns to create fake installs with low quality users.
What Is Mobile App Cost Per Install (CPI) Fraud?
Due to the ever growing popularity of mobile devices like smartphones and tablets, mobile advertising has recently experienced a significant amount of growth. Since more and more money is flowing into the mobile advertising industry, this industry has attracted more malicious activities than ever before. After all, cost per install (CPI) payouts can range anywhere from $1 to $10, making the potential for profit very high for cybercriminals. In turn, this means that the potential for losses is equally as high for businesses and advertisers.
From 2015 to 2018, multiple analytics companies estimated that billions of U.S. dollars had been lost to fraud in the mobile advertising industry. We estimate that around $2 billion was lost in 2017 specifically to mobile app install fraud. One particularly common example of malicious activity in this industry is cost per install fraud.
What Is Cost Per Install (CPI)?
A cost per install (CPI) campaign is mainly designed for mobile platforms, but can extend to some desktop software as well. In a CPI campaign, publishers have digital advertisements placed across many different media platforms to encourage the installation of the mobile application being advertised. The brand is charged the bid or fixed rate only when a user installs and opens the application.
What Is Cost Per Install Fraud?
CPI fraud involves the creation of fake installs to deplete the budget of the advertiser or tap into a lucrative affiliate install campaign that would allow a cybercriminal to generate easy profits in the form of commissions. There can be hundreds or even thousands of fake installs generated per day for the most popular CPI campaigns. In general, bots, malware, and proxies use automated scripts to generate these fake installs. Other forms of cost per install fraud include the following:
- Bots (a proxy generates automated installs)
- Device emulators (simulated devices generate installs and these devices are able to send signals that appear genuine)
- Click farms (downloads made by real individuals who own multiple devices and receive payment for each download)
- Maliciously incentivized traffic
- Click spam—also referred to as click stuffing—is the practice of attributing the installation of an app to a fake source. This is usually done through the use of a fake app developed by the cybercriminal. Click spam is becoming one of the most common forms of fraud when it comes to CPI.
The vast majority of the routes taken for CPI fraud can be used for other models, especially other types of digital ad fraud.
How to Fight CPI Fraud?
Now that you know what mobile app CPI fraud is, chances are you're wondering how you can fight this phenomenon.
Smart Server Farms / Install Bots
Bots refer to software that is able to imitate human behavior. In this case, bots automate the installation of an app. In some cases, bots are even able to re-install apps if they have already been installed.
The best way that you can tackle this issue is by linking each installation with a unique identifier of the device. This identifier is referred to the Device ID and each mobile device generally only possesses one such identifier. This will prevent CPI fraud involving multiple installations from the same device. Unfortunately, if you're dealing with someone who is a master at fraud, they may create different IDs from the same mobile device. One sign that you're dealing with CPI fraud involving bots is a large number of installs from a single IP address or strange timestamps.
IPQ's Device Fingerprinting technology is the answer to detecting Device ID fraud. We use over 200 data points to determine if there is any strange behavior related to a user's online fingerprint.
Analyze Behavior After Installation
One of the best ways that you can identify CPI fraud is by analyzing the behavior of the user after installation. The higher you go in terms of the conversion funnel, the more difficult it will be for cybercriminals to fake human behavior. Studying post install data will make it easy for you to identify fraudulent patterns.
For example, if you are seeing some users coming from a single source with no in-app activities after installation, you should flag such users as fraudulent. In-app activities can include something as simple as clicks within the app. You want to focus on the users who enjoy and use your mobile. To maximize CPI payouts, focus on what your users do after downloading. Not only do you want to know if they're actually using your mobile app, but you also want to know about their behavior.
Machine Learning Algorithms
Manually, to identify CPI fraud, you can detect traffic behavior changes like:
- Major changes in the distribution of types of devices
- Click to install ratio to detect click flooding and install hijacking
- CR metrics for signs of maliciously incentivized traffic
- Intervals between actions are too suspicious (short, even, etc.)
As you can probably imagine, fraud detection is not a simple task. This is particularly true if you have a daily volume of thousands of installs each day. Therefore, you need to think about either hiring a team to help you or using software—like ours—for fraud detection. Today, many types of fraud detection software rely on machine learning algorithms to help detect signs of CPI fraud.
One of the main issues with ad fraud is its diversity as well as its ability to adapt to changes in the advertising industry. Fortunately, experts are learning more and more about how to identify and fight CPI fraud and other forms of ad fraud. If you would like to learn more about how to protect yourself against mobile app CPI fraud and other forms of mobile app fraud, don't hesitate to contact us.
