User Guide

Anti-Fraud Metrics

Anti-fraud from AdsBridge is a complex algorithm based on artificial intelligence. This technology analyses traffic following several metrics in common. Traffic will be counted as the fraud only when metrics show a high probability of this event as a fraud.

Thanks to such kind of traffic analysis anti-fraud detection on AdsBridge gives a clear picture of fraud events.

METRICS TO DETECT THE FRAUD TRAFFIC ON ADSBRIDGE:

Frequent events 

If specific IP-address generates more than 60 events per second which is more than the general user may perform, then the possibility of fraud activity is enormously high. That’s why such visits/clicks/conversions are counted as a fraud.

 

Data centered 

Some widespread data centers and cloud platform providers might be a fraud traffic source. It doesn’t mean that every visit from the IP address associated with the data is not valid (for example, VPN users can be routed through data centers), but it increases the chance to be the non-human origin.

 

Fake timezone 

The timezone was falsified by other software

 

Misleading language 

When the language doesn’t fit the geo-location, it possibly means that the event is a fraud. 

 

Unrecognized devices

Incoming traffic shares info about the device from which the redirect was made, and the device type, manufacturer and model are not recognized. It means that traffic is a fraud.

 

Fast clickers

Fraud traffic doesn’t have time delays between ad performance and click on this ad, otherwise, visits from real users show specific time difference between it. So, if the time interval between the display and the click is shorter than 1 second, such an event is likely to have a non-human origin.

 

Library robots 

This metric uses User-Agent information to determine unwanted traffic to your site. If User-Agent says that this is a library, then the probability that this transition is fraud appears to be very high. These libraries contain the User-Agent spam bots, blacklist bots, scanning bots and tools that can harm your website’s content. This bot can simply scan the content of your site (for example, Google search robots) and sometimes might not belong to fraud.

 

Unrecognized user agent 

This metric uses information about the User-Agent. If redirect uses wrong User-Agent format or is empty, this event is mostly like to be a fraud.

 

Without referer 

Referrer contains information about the source/site the redirect was made. If the referrer is absent, then there is a probability that visit appears to be the non-human origin.

 

Unsupported OS versions 

If the user’s operating system is recognized as an old and doesn’t support for a long time, it is possibly a fraud.