Maximizing User Acquisition ROI with AI Technology

In recent years, many mobile applications, including mobile games, have expanded to the global market. However, as they expand, fake traffic is becoming a growing problem that has long plagued many game developers. It directly results in the waste of marketing resources, making it impossible for developers to have real control over the ROIs of acquiring new users.

Fake Traffic – A Growing Plague to Marketing ROI

Mobile advertising has developed at a high speed, but the problems of fake traffic and fake installations have reached a point where they are difficult to handle. Fake traffic not only directly encroaches on the marketing expenses of advertisers, but also causes the traffic value of high-quality channels to be inaccurately evaluated. The cost per installation has gradually increased from $1 to $10, and app promotion has grown to a huge market of over $3 billion. On average, 10-20% of paying users in each application are allotted budgets that are fraudulently misappropriated. This number is more than 50% for some channels.

Evolving Fraud Methods

Fake traffic has affected the entire industry. For example, many industry giants such as Facebook and Google are equally annoyed: Facebook shut down its own DSP platform because of fraudulent traffic, and Google has generated a large number of refunds due to fake traffic.

Various methods can be used to generate fake installations. Fraudsters will use apps to hijack devices at the right time and create seemingly legitimate “ad clicks”, allowing them to earn installation (CPI) payments. While monitoring global mobile app promotion, we found the following means used to create fake traffic:

Mobile device emulators: Emulate large numbers of different mobile devices on one device
Fake installation factory: Manual app installation
Malicious apps: Installing additional apps without the user’s permission
Leveraging cloud services: Creating a large number of virtual users at different IP addresses
Anonymous agents: Using anonymous agents to install an app in one country, but using the app in another country

These are just a few of the methods used, and fraudsters will use complex tactics to evade fraud detection and pose as normal users. They can no longer be easily handled by simple detection rules. The industry urgently needs a more advanced technology to solve this problem.

DataVisor Uses AI to Bring Trust Back to the Game Ecosystem

DataVisor’s unsupervised machine learning engine is proven in the industry with production grade scalability. Traditional methods of detection often rely on a rules system or a model that requires training data, both of which are inadequate against modern fraud methods. DataVisor automatically and accurately detects changing fraud patterns without the need for training data. DataVisor’s algorithm is based on the latest Spark big data system. It can cluster and correlate user behaviors and find hidden, abnormal associations among a large number of users. At present, DataVisor has worked with many large game companies such as IGG, Funplus, Cheetah Mobile, Elex, etc. and has helped them expand their user bases and enhance the user acquisition ROI.

DataVisor helps IGG save millions of dollars in fraudulent user install fees each year and ensures the healthy growth of their user community. DataVisor allows companies to invest more confidently while being free from the hassles and losses caused by fraudulent installations.