How it Works

If you want to get more technical

From an internet user’s point of view, Adscore will first collect technical information about their browser and device using JavaScript. This includes details like the screen resolution, or the browser’s language. The data is then compressed and sent to one of our servers where the actual analysis and detection takes place. Data is then stored only in-memory and only for the duration of the analysis itself (about 30-50ms). In order to respect and ensure the privacy of internet users, no detailed data gathered from the visiting browser is kept for longer than necessary.

Adscore uses multiple technologies in order to detect fraudulent web traffic, that actually strengthen each other.

First of all, we’ve got over 100 hand-crafted detection algorithms covering a broad spectrum of traffic, that are based on our decade-long experience — as an example, detecting mobile devices that don’t have a touchscreen.

Secondly, all traffic passes through our distributed machine learning system, that automatically learns how do each browser and device behave. It then uses that knowledge to validate the traffic. This mechanism can, for example, learn the screen resolution of a particular mobile device model and cross checks it against traffic it will see in the future. It’s also able to monitor IP address ranges or Internet Service Providers, tagging them as suspicious depending on certain statistical analysis processes.

We keep a close eye on our detection mechanisms. Every day we monitor and improve them, by manually analyzing parts of the traffic and verifying the results of the various modules that make up Adscore.