Filter High Risk Traffic

impact.com monitors clicks for specific risk signals that indicate they originated from a non-human source. While these clicks can be either innocent or malicious, they can still distort your performance metrics and result in unwanted payouts. impact.com allows you to set how strict you want the click filter policy to be for your program.

1. From the top navigation bar, select [User profile] → Settings.
2. In the Program Settings column, scroll to the Scoring heading and select Event Scoring & Filtering.
3. Select either Relaxed or Moderate from the [Drop-down menu]. See Filtering Options below for more details.

   • Contact support to enable the Strict option.
4. Select Save.

Filtering Options

There are 3 filter settings that can be applied to clicks displaying high-risk signals. The table below provides a handy reference for the different filtering options:

Click Filter Setting	Description	Included in Performance Reporting	Eligible for CPC	Eligible for attributing actions
Relaxed	Clicks are analyzed for high-risk signals but impact.com won’t act on them.	Y	Y	Y
Moderate	Clicks are filtered from performance reports, are ineligible for CPC, but will be able to attribute actions. This is the recommended setting to protect the integrity of your program metrics and prevent the most vulnerable payout events, while still allowing you to review actions attributed from these clicks and decide if they should be paid out.	N	N	Y
Strict	Clicks are filtered from reporting, are ineligible for CPC, and cannot attribute actions.	N	N	N

Below are the 3 high-risk signals that impact.com identifies.

Data center

Description	Why do I see this in my program?	False Positive chance
The click originated from an IP address that belongs to a data center like AWS or Google Cloud.	Clicks from data centers represent non-human traffic, which can be innocent, like web scrapers or partners inadvertently buying bad traffic, or malicious attempts to defraud your program.	Very low, occasionally caused by: • Partners using an unsupported implementation. • IP ownership changing and enrichment services holding out-of-date information.

HTTP Request Anomaly

Description	Why do I see this in my program?	False Positive chance
The http request contains signals that indicate non-human activity.	Similar to data center clicks, these HTTP request anomalies can be malicious intent or innocent, but are still non-human in nature.	Very low, occasionally caused by partners using an unsupported implementation.

Proxy

Description	Why do I see this in my program?	False Positive chance
The IP Address that the click originated from is a known proxy.	Clicks from known proxies are likely to be malicious in intent since illegitimate actors often use proxy IPs to mask invalid traffic. Events flagged for this reason should be carefully reviewed before approving payouts.	Very low.