The way in which test traffic is split, or allocated.

Equal allocation
Typically, in an A/B test, traffic is equally allocated so half of visitors see one version and the other half see the variant. This equal 50/50 split is known as equal traffic allocation.

In an A/B test with a control and more than one variant, traffic can still be allocated equally if for example visitors are split 33/33/33 across a 3 total of versions.

Dividing traffic equally, is the fastest, most accurate, and most reliable way to conduct an A/B test.

Unequal allocation
However, it is possible to split traffic other ways, for example 80/20 or 70/30.

This splitting is known as unequal allocation.

It occurs when a smaller or greater proportion of users are allocated to either the control or the variants.

As this article explains, unequal traffic allocation it is not recommended. When traffic is not equally allocated, the results tend to be more uncertain and may take longer to reach a significant conclusion.

Therefore, equal allocation, or a 50/50 split is the optimal traffic allocation.