We have implemented a new approach to uncovering fraud that is hard to detect through regular means (like CTIT). It relies on identifying ad traffic anomalies using Big Data and machine learning.
AI Fraud Metric is a new fraud detection mechanism in Fraud Scanner that uses machine learning models. These models are based on data gathered across ad interactions, installs, launches, sessions, payments, and in-app user activities.
Fraud Scanner now automatically generates multiple cohorts and singles out those that have anomalous distributions compared to others. This makes it possible to identify complex and unconventional types of fraud. An example would be “short-burst” fraud, where clean traffic is contaminated with small amounts of fraudulent installs.
AI Fraud Metric helps detect well-disguised fraud that is hidden in quality traffic. This is especially important for large-scale campaigns with hundreds of ad placements. In cases like this, fraudsters usually go for one placement as opposed to all of them. Manually searching for fraud in such a campaign would be a Herculean task.
If you are registered, simply head to our Report Builder and add the new AI Fraud Metric as shown on the screenshot below:
If you are not yet registered, follow these three easy steps to become one. After that, follow the above instruction to add the new metric in the Report Builder.
The new AI Fraud Detector will be available to all users in the next 30 days, after this period of time it will only be available to Premium users.
Let our ML models do the legwork for you, providing fast and accurate fraud detection. Be sure to try it while it's free!
