IP filtering allows you to exclude specific IP addresses from your experiment data. This is particularly useful for excluding internal traffic that could skew your experiment results.
Note: IP filtering only affects experiment data collection. Users with excluded IP addresses will still see the experiment variants, but their data will not be included in your experiment results.
How to Set Up IP Filtering
To set up IP filtering for your project:
Go to your domain settings.
Click on "View Settings" in the Domain header.
Scroll down to the "IP Address Exclusion" section.
Enter the IP address you want to exclude in the input field and click "Add".
Repeat for any additional IP addresses you want to exclude.
When to Use IP Filtering
IP filtering is particularly useful in the following scenarios:
Excluding internal traffic: If you and your team frequently visit the site, you might want to exclude your office IP addresses to prevent your own behavior from affecting the experiment results.
Excluding development environments: If you're running experiments on a development or staging environment, you might want to exclude those IP addresses.
Excluding known bots or crawlers: If you know certain IP addresses belong to bots or crawlers, you can exclude them to ensure your data only includes real user behavior. But we already have filtering logic for these as default.
IP Address Format
IP addresses must be entered in the standard IPv4 format: xxx.xxx.xxx.xxx, where each xxx is a number between 0 and 255.
Examples of valid IP addresses:
192.168.1.1
10.0.0.1
172.16.254.1
Limitations
There are a few limitations to keep in mind when using IP filtering:
IPv6 not supported: Currently, only IPv4 addresses are supported.
No wildcards: You cannot use wildcards or ranges to exclude multiple IP addresses at once. Each IP address must be added individually.
No subnets: You cannot exclude entire subnets or IP ranges. Only individual IP addresses can be excluded.
Tip: If you're using a VPN or proxy, make sure to exclude those IP addresses as well to ensure your own traffic is not included in your experiment data.