Fraud is a functionality that defines inappropriate traffic. You can find the metrics related to this functionality in the statistics table (bottom left corner, gear icon):
Fraud metrics show how much of your traffic is considered suspicious. Suspicious traffic is divided into two categories:
- Suspicious visits. Shows the count and the percentage of visits marked with at least one indicator as suspicious.
- Suspicious clicks. Shows the count and the percentage of clicks marked with at least one indicator as suspicious.
Clicks are only visible if you are using landers in your campaign funnels.
To find out why the traffic was identified as suspicious:
- Select a campaign.
- Click the Fraud details button on the control panel.
A report will open, where you can find detailed information, according to which metric visits and clicks are considered suspicious.
Setting Up Fraud Metrics
Configure metrics to check incoming traffic for suspicious visits and clicks:
- Go to Settings.
- Go to the Fraud section.
- Select the checkboxes for the metrics you want to use to check your traffic.
- Click Save.
From the moment you change the settings, your traffic is checked according to the newly set metrics. The past data will not be recalculated according to the newly installed settings.
Review each metric below and determine which traffic is unacceptable in your funnels.
Database bots and spies
Own and third-party bases of bots and spies.
VPN and Proxy
Traffic using VPN / Proxy connections. People often use this technology to hide their location, bypass the lock, and are on the Internet under their IP address.
Registration of visits with an empty / non-transmitted referrer.
A metric that marks traffic if the user's device is not recognized.
PeerClick uses special software to determine if a visit or click came from a known datacenter by IP address. Data centers such as Amazon Web Services have their own IP addresses assigned to them. Some of the datacenters may be the source of inappropriate traffic, but keep in mind that VPN users can route through the datacenters. Thus, a visit from an IP address associated with the datacenter does not necessarily mean that it is invalid, but it raises that likelihood.
A metric that marks traffic if the user agent class claims to be a library.
Internet Protocol version 6 traffic. At the moment, it is poorly distributed in the CIS and is not used by private individuals, which increases the likelihood of such traffic being a bot traffic.
Unrecognized user agent
Metric marking traffic with an unrecognized user agent. User agent information must be transmitted in the correct format, and most of the traffic follows this procedure.
Unsupported OS version
Information about the visitor’s operating system and its version can also be obtained from the user agent. If the system is recognized as old or not recognized at all, it is most likely a bot.
Invalid location language
Mismatch of the country and language of the browser / user's device in one event.
A metric that marks traffic if a single IP generates more than 60 events per minute.
Frequent campaign events
A metric that marks traffic if a single IP generates more than 60 events per minute per campaign.