# A/B Testing
## Creating an A/B Testing
A/B testing is automatically enabled when you add more than one notification variant. To create an A/B test:
1. From the Create Notification page, locate the **Content** section. Your initial notification appears under the **Notification** tab.
2. Click the green **+** button next to the variant tab. A dropdown appears with two options:
* **Add Blank Test Variant** — Creates a new empty variant that you can configure from scratch.
* **Duplicate Notification** — Creates a new variant pre-filled with the content from the current notification.
3. Once a second variant is added, the tabs are automatically renamed to **Variant A**, **Variant B**, and so on.
4. Click on each variant tab to configure its content independently — you can vary the title, body, image, landing URL, or any other content field.
⚠️ Test naming and traffic distribution options appear at the end of the form when multiple variants are added.
### Managing Variants
Each variant tab has a **three-dot menu** ( **⋮** ) that provides options for managing that variant.
When multiple variants exist, the menu for each variant tab contains:
* **Duplicate** — Creates a copy of this variant as a new tab.
* **Reset content** — Clears all content fields in this variant and returns them to their default empty state.
* **Remove** — Deletes this variant from the test.
* **Remove All Other Test Variants** — Removes all other variants, keeping only the selected one and returning to a single notification (non-test) state.
### Test Name
Naming your test is important for future identification of notifications that comprise a test as well as identifying performance for reporting and analytics purposes within the platform.
{% hint style="info" %}
A test name cannot be used more than once. We suggest adding dates to test names if you want to a single test multiple times.
For example: `Image Test / 2020-02-10`
{% endhint %}
## Variant Distribution
Once you've created each variant you want to test, the `Test Distribution` section on the top right- hand side of the page provides options that allow you to determine how your test will be deployed to your audience.
### Split Delivery
Split Delivery allows you to choose a percentage of your audience that each variant should receive. This distribution type allows you to test new variants on small percentages of your audience while ensuring clickthrough rates stay high with what you already know works.
### Sample Delivery
Sample Delivery delivers a sampling percentage of your audience to each variant before deploying the best performing notification to the rest of your audience.
Use the sliders to determine what percentage of your audience each variant and the winner will receive. Each variant must receive at least 5% of the audience's delivery.
When Sample Distribution is chosen a `Winner Strategy` section appears that allows you determine when the variant with the best performance is deployed to the rest of your audience. There are two ways to configure this timing:
1. **Time Elapsed After Send:** Specify the number of hours that have elapsed after the initial variants were sent.
2. **Specific Time:** Schedule the winning notification to be delivered at a specific date and time in the future.
{% hint style="warning" %}
This option is not available for audiences with fewer than 5,000 subscribers.
{% endhint %}
#### Performance Constraint (Sample Delivery Only)
When running a Sample Delivery test, you have the option to activate a performance constraint. If none of your deployed variants in the A/B Test reach the CTR value placed in the performance constraint, then no Notification will be deployed to the remaining audience.
---
# Agent Instructions: Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter:
```
GET https://documentation.pushly.com/platform/notifications/a-b-testing.md?ask=
```
The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
