# What traffic allocation for an experiment on a niche audience?

Asked by Diego Herrera on 2026-05-12. Tags: ab-testing, audiences, methodology.

We run ABM personalization and I want to experiment on an audience of
enterprise accounts that represents roughly 4% of total traffic. The general
advice I keep seeing is to allocate around 30% of traffic to an experiment
to limit risk.

Does that guidance apply here? If allocation is a percentage of the eligible
audience rather than of total traffic, 30% of an already tiny segment seems
like it would take months to reach anything conclusive. What is the correct
setting for this case

## 1 answer

### Accepted answer from petra_v (2026-05-12)

Your reading is right: traffic allocation is the percentage of the
ELIGIBLE audience that gets entered into the experiment. If you set 30%,
then 30% of the visitors matching your enterprise audience see variants
and the remaining 70% see the default content. It is not a percentage of
overall site traffic.

For a niche audience like your 4% segment, use 100% allocation. Every
variant needs to reach at least 1000 visitors and 25 conversions before a
winner can be called, and if you feed the experiment only a third of an
already small segment, hitting those thresholds takes prohibitively long.

Partial allocation exists mainly to de-risk experiments on high-traffic
pages, where you do not want a bad variant exposed to everyone at once.
In your case the audience definition is already the risk limiter: only 4%
of visitors can ever see the experiment, so there is little left to hedge.