For the complete documentation index, see llms.txt. This page is also available as Markdown.

Traffic Allocation

Use the Traffic allocation step to define how the campaign traffic is split between the Original and your variation.

In Feature Experimentation and Rollout, this step uses a standard traffic allocation for A/B tests. You decide what percentage of the campaign traffic goes to the variation, while the remaining tracked traffic stays on the Original.

How it works

The page is divided into two main areas:

  • Global traffic: where you define how much traffic is allocated to variations, how much remains on the Original, and how much is left as untracked traffic

  • Traffic distribution by variation: which shows a visual summary of the current split

For a simple A/B test with one variation:

  • the traffic allocated to the variation is taken from the total campaign traffic

  • the remaining tracked traffic is allocated to the Original

  • untracked traffic is not included in the test results

If untracked traffic is set to 0%, all campaign traffic is tracked and split between the Original and the variation.

Configure the traffic allocation step

  1. In Global traffic, set Traffic allocated to variations.

  2. Review the Original value, which adjusts based on the remaining tracked traffic.

  3. In Variations Traffic, review the allocation for your variation.

  4. Check the chart in Traffic distribution by variation to confirm the final split.

  5. Click Save.

For a standard A/B test with one Original and one variation, the recommended setup is:

  • 50% for the Original

  • 50% for the variation

  • 0% untracked traffic

An even split is usually the simplest setup for comparing performance between the two versions.

Important details

  • Untracked traffic sees the Original version but is not counted in the test reporting.

  • Lowering tracked traffic also lowers the size of the test sample.

  • If you change the traffic allocation while the campaign is live, only new visitors are affected by the new split.

  • To keep reporting easier to interpret, it is best to avoid changing the allocation during the campaign unless necessary.

Last updated

Was this helpful?