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

# Create an A/B Test

An A/B test splits a fixed share of your traffic across a control group and up to three variants, then measures which one performs best. This guide walks the full flow, from the create button to a live experiment.

## Prerequisites

* Permission to manage experiments in the workspace.
* At least one channel and language configured.
* The pages, components, algorithms, or merchandising rules you want to test, each with variants already defined.

## Create the Experiment

<Steps>
  <Step title="Start the flow">
    On the Experiments list, click **Create Experiment**, then choose **A/B Test** in the Select Experiment Type dialog.

    <Frame>
      <img src="https://mintcdn.com/experro/vL9u48aLzyWUbwlI/images/experiments/a_b_testing/imagesource_createexperiment.png?fit=max&auto=format&n=vL9u48aLzyWUbwlI&q=85&s=9eb746ee95f26601214330cf072b987f" alt="" width="1440" height="900" data-path="images/experiments/a_b_testing/imagesource_createexperiment.png" />
    </Frame>
  </Step>

  <Step title="Enter the details">
    Give the experiment a name, then select a **Channel** and a **Language**. Observation and hypothesis are optional but worth writing — they record why you ran the test.

    Click **Create & Continue**. This saves the experiment as a draft.
  </Step>

  <Step title="Define your metrics">
    Add one primary metric. This is the KPI that decides the winner. Add secondary metrics to watch for side effects.
  </Step>

  <Step title="Choose your audience">
    Decide who enters the experiment and what share of them take part.
  </Step>

  <Step title="Set the duration">
    Pick a start date. Optionally add an end date, an event count limit, and email recipients to notify when the test finishes.
  </Step>

  <Step title="Allocate traffic to variants">
    Add your variants and set the split. The total must equal 100%.
  </Step>

  <Step title="Map the experiences">
    Add the pages, components, algorithms, or rules you are testing, then map each one to a variant.
  </Step>

  <Step title="Launch">
    Click **Start Experiment**, then confirm the prompt to publish the associated records.
  </Step>
</Steps>

<Note>
  **Start Experiment** stays disabled until you complete the Details step. Its tooltip reads "To start the experiment, first add the experiment details and create it."
</Note>

## If Validation Fails

Experro checks every step when you start the experiment. If something is missing, a red icon appears beside the affected step in the **Configurations** rail and the message "Please enter all the required fields" appears at the top of the screen.

Use the rail as a checklist. A green check means the step is complete.

<Tip>
  You can click **Save** at any point, even with required fields empty. Saving keeps the experiment as a draft so you can come back to it.
</Tip>

## Key Considerations

* After you save the Details step, **Channel** and **Language** become read-only.
* Every experience you add must be mapped to a variant before the experiment can start.

## What's Next

* [Details](/experiments/a_b_testing/details)
* [Metrics](/experiments/a_b_testing/metrics)
* [View Experiment Results](/experiments/analytics/overview)