How traffic split works

How traffic split works

When a visitor lands on a page with an experiment, Netaj uses a deterministic hash of their visitor ID + the experiment ID to pick a variant. Same visitor on the same experiment always lands in the same bucket — no flicker, no "they saw both".

Weighted splits

By default, traffic is split evenly across variants. You can weight variants differently — for example, 80% control / 20% test if the change is risky and you want a slow rollout.

Targeting

An experiment only affects visitors that match its targeting rules: URL pattern, device type, country, or custom attributes. Visitors outside the targeting just see the control.

Multiple experiments on the same page

If two experiments target the same page, each visitor is randomly assigned to one of them — never both. This avoids confounding results.