> For the complete documentation index, see [llms.txt](https://docs.shoplift.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.shoplift.ai/api-reference/public-api-1/shoplift-api/overview.md).

# Overview

## What you can do with the Shoplift API

The Shoplift API is a REST API that lets you create and manage A/B tests, custom audiences, and campaigns on your store programmatically. Anything you can do with tests in the Shoplift dashboard, you can do with an API call. The API works with any tool that can make HTTP requests: scripts, internal dashboards, automation platforms, and AI assistants like Claude.

The API is available on all Shoplift plans.

### What you can do

**Create and manage tests.** Create template, theme, URL redirect, and API tests in draft state, then launch, pause, resume, or end them. Update drafts freely, and update titles and hypotheses even after launch.

**Pull live results.** Retrieve real-time statistics for any launched test: visitors, conversions, revenue, lift, and Bayesian win probabilities, broken down by device and visitor type, with daily history.

**Organize your testing program.** Tag tests, set a priority (Low, Medium, High, Critical), score expected impact and effort on a 1 to 5 scale, and track whether each test's winning variant was implemented. Shoplift sets the implemented flag automatically when a test ends with its winning variant launched through Shoplift. If you implement a winner outside Shoplift, set the flag yourself via the API. These fields turn your test list into a prioritized testing roadmap you can manage programmatically.

**Target and schedule.** Build reusable custom audiences from country, channel, UTM, URL, and referrer rules. Schedule tests to start and end automatically, or end them automatically when they reach statistical significance (the point at which a result is unlikely to be down to chance).

**Work across campaigns.** Associate tests with active campaigns to keep larger initiatives organized.

{% hint style="info" %}
Campaigns require a plan that includes the [Campaigns feature](/test/campaigns/overview.md). 
{% endhint %}

### Common use cases

* **Custom reporting**: combine Shoplift test results with your other business data in your own dashboards and BI tools
* **Automated monitoring**: get alerted in Slack when a test needs attention, such as traffic split issues, low volume, or a variant trending badly
* **Scheduled reporting**: deliver test summaries automatically at key milestones, such as 7 days live, 14 days live, or test ended
* **Testing roadmap operations**: keep a live view of your program, including which winners haven't been implemented and what's next by priority, impact, and effort

Step-by-step recipes for these live in [Guides](/api-reference/public-api-1/shoplift-api/api-guides.md).

### Do I need an MCP server?

No. Shoplift doesn't ship an MCP server because the API is designed so AI assistants can use it directly, and in practice that gives you more, not less.

An MCP server is a wrapper: it translates an API into a fixed set of tools an AI assistant can call. That works well when an API is sprawling or poorly documented. The Shoplift API is small, consistent, and fully described in a single reference. Given that reference and your API key, an assistant like Claude can call every endpoint directly, including new ones the moment they ship, with no wrapper to install, update, or wait on.

Working directly with the API also means:

* **It works everywhere.** Claude, Claude Code, Cursor, custom agents, and plain scripts all speak HTTP. You aren't limited to tools that support a particular MCP client.
* **Full surface area, day one.** Every field and endpoint in the reference is available immediately. There's no lag between an API update and a corresponding tool update.
* **One credential, one mental model.** The same API key and the same reference documentation serve your developers, your automations, and your AI workflows.

If you want the "connect and go" experience an MCP promises, see [Use the Shoplift API with Claude](/api-reference/public-api-1/shoplift-api/shoplift-api-with-claude.md). Setup takes about two minutes.

### Authentication, limits, and scope

* Every request is authenticated with a Bearer token: your store's API key, which starts with `sl_live_`. See [Get and manage your API key](/api-reference/public-api-1/shoplift-api/shoplift-api-key.md).
* Keys are store-scoped. One key grants access to one store, and each store has one active key.
* Requests are rate limited to 60 requests per minute.
* Personalization tests are not accessible through the API.

Ready to build? Head to the [API reference](https://docs.shoplift.ai/api-reference) for full endpoint documentation, or start with [Use the Shoplift API with Claude](/api-reference/public-api-1/shoplift-api/shoplift-api-with-claude.md).

### Still need help?

Contact the Shoplift support team if you have questions about the Shoplift API.