> ## Documentation Index
> Fetch the complete documentation index at: https://confidence.spotify.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Audience

> How the audience works in an experiment: who is eligible, what share of them you allocate, how Confidence randomizes them, and how sticky assignment and exclusivity apply.

The audience of an experiment is the set of users or other units that you make eligible for it. Experiments use the same targeting model as flag rules and segments. This page covers how that model applies inside an experiment. For the full reference on criteria, allocation, randomization, and sticky assignment, see [Audience targeting](/docs/capabilities/segments/audience).

You set the audience in the **Audience** section of an A/B test or rollout. It has four parts:

* **Inclusion criteria** decide which units are eligible for the experiment.
* **Allocation** sets the percentage of the included population to run the experiment on.
* **Targeting key** specifies the field in the evaluation context that Confidence randomizes traffic on.
* **Exclusivity** controls how the experiment relates to other experiments, rules, and segments.

## Inclusion Criteria

You can combine these criteria types:

* **Attributes**: Target users based on evaluation context fields like country, platform, or version.
* **Segments**: Target users who belong to a pre-defined [segment](/docs/capabilities/segments/overview), so you reuse the same audience across experiments.
* **Holdbacks**: Include or exclude a random subset of users defined by a [holdback](/docs/capabilities/surfaces/surface-settings#holdbacks) on a surface.
* **Groups**: Combine multiple criteria with `AND` or `OR` operators to build complex targeting logic.

To configure criteria, follow [Define the Audience](/docs/capabilities/segments/define-audience-criteria).

## Allocation

You can increase the allocation of a live A/B test, but you can't decrease it. For rollouts, you control the allocation through the [rollout reach](/docs/guides/roll-out-a-change). Read more in [Allocation](/docs/capabilities/segments/audience#allocation).

## Randomization

Confidence randomizes units into variants using the value of the targeting key field in the evaluation context. Choose the field that maps best to the experience you are testing. Read more in [Randomization](/docs/capabilities/segments/audience#randomization).

## Sticky Assignment

To enable sticky assignment, open **Advanced options** in the **Audience** section. With sticky assignment on, Confidence stores every assignment so that units keep their variant, and you can pause intake of new units. Read more in [Sticky assignments](/docs/capabilities/segments/audience#sticky-assignments).

## Exclusivity

To keep units from being in two experiments at the same time, add the experiments to the same exclusivity group. Exclusivity groups live on surfaces. Read more in [Exclusive experiments](/docs/capabilities/experiments/exclusive-experiments) and [exclusivity groups](/docs/capabilities/surfaces/surface-settings#exclusivity-groups).

## Related Resources

<CardGroup cols={2}>
  <Card title="Audience Targeting" href="/docs/capabilities/segments/audience">
    Full reference for audience configuration
  </Card>

  <Card title="Define Audience Criteria" href="/docs/capabilities/segments/define-audience-criteria">
    Configure targeting criteria
  </Card>

  <Card title="Exclusive Experiments" href="/docs/capabilities/experiments/exclusive-experiments">
    Set up mutual exclusion
  </Card>

  <Card title="Treatments Reference" href="/docs/capabilities/experiments/treatments">
    Configure experiment variants
  </Card>
</CardGroup>