Google Analytics MCP Server
Connect Claude, ChatGPT, or any AI tool to your Google Analytics 4 properties. Ask questions about your website traffic, conversions, and audiences in plain English — no GA4 report-builder required.
What is the Google Analytics MCP?
The Pipeboard Google Analytics MCP (Model Context Protocol) Server connects AI assistants like Claude, ChatGPT, and custom agents to the Google Analytics 4 (GA4) Data and Admin APIs. Instead of building reports by hand in the GA4 interface, you ask questions in plain language and the assistant pulls the answer directly from your Analytics properties.
It is built for marketers, analysts, and agencies who want fast answers about website and app performance — sessions, users, traffic sources, conversions, engagement, and revenue — without learning the GA4 report builder or the underlying API.
What you can do with it:
How Pipeboard uses your Google Analytics data
What we request, why, and what we never do with it
When you connect Google Analytics, Pipeboard requests access through Google’s official OAuth consent screen. We only request the Google Analytics scopes needed to run the reporting and configuration tools described on this page, plus your basic account identity so we can show which Google account a property belongs to. We use this access solely to fulfil the requests you (or your AI assistant, on your behalf) make.
The access we request, and why:
analytics.readonlyRuns reports against your GA4 properties — sessions, users, conversions, revenue, traffic sources, landing pages, realtime activity — and lists the properties, dimensions, metrics, and conversion events available to you. This is the read access that powers every reporting tool.
analytics.editUsed only when you explicitly ask the assistant to change a property setup — creating or updating custom dimensions and custom metrics, or creating and deleting conversion events. It is never used to alter or delete your historical analytics data, which GA4 does not allow in the first place.
userinfo.emailIdentifies the connected account so we can label which Google login a property belongs to. Important if you connect more than one Google account.
userinfo.profileDisplays the connected account name in your Pipeboard connections list so you can tell your connections apart.
Our commitments
- We never sell your Google Analytics data, and we never use it for advertising.
- Your data is used only to respond to the requests you or your AI assistant make — it is not used to train AI models.
- You can disconnect Google Analytics at any time from your connections page, which revokes Pipeboard’s access.
- Read-only by default: the edit scope is only exercised when you explicitly ask to change a property’s configuration.
- Pipeboard’s use of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.
Full details are in our Privacy Policy and Security pages.
Getting Started
Set up in under 2 minutes
1. Sign up for a Pipeboard account:
pipeboard.co/auth/signup →2. Connect your Google Analytics account (you’ll review and approve the access above on Google’s consent screen):
pipeboard.co/connections →3. Add the MCP server to Claude (or another MCP client):
https://google-analytics.mcp.pipeboard.co/Use this URL as a Streamable HTTP MCP server in your client. Authentication is handled automatically through your Pipeboard session.
Connect to Your AI Tool
The Google Analytics MCP works with any MCP-compatible client. Just add the server URL and start asking about your analytics in natural language.
Claude Pro / Max
Recommended- 1.Open claude.ai/customize/connectors
- 2.Click "Add custom connector"
- 3.URL:
https://google-analytics.mcp.pipeboard.co/
Claude Code
CLI / VS CodeRun this command in your terminal:
claude mcp add --transport http pipeboard-google-analytics https://google-analytics.mcp.pipeboard.co/Then type /mcp in Claude Code to authenticate with Pipeboard.
Alternative: use an API token for instant setup (no browser auth needed):
claude mcp add --transport http pipeboard-google-analytics https://google-analytics.mcp.pipeboard.co/?token=pk_...Get your token at pipeboard.co/api-tokens
Claude Desktop JSON configuration:
Add this to your Claude Desktop MCP settings (Settings → MCP Servers → Add):
{
"mcpServers": {
"google-analytics": {
"url": "https://google-analytics.mcp.pipeboard.co/"
}
}
}More integrations:
Try these prompts
Traffic Overview
"List my Google Analytics properties and show sessions, users, and conversions for the last 28 days"
Channel Comparison
"Compare traffic by default channel grouping this month vs last month and tell me what changed"
Landing Pages
"Show my top landing pages by sessions, with engagement rate and conversions, for the last 30 days"
Live Activity
"How many active users are on my site right now, and which pages are they on?"
What’s included
Reporting (read-only)
- Standard, pivot, and batched GA4 reports
- Realtime reports for live activity
- Property, dimension, and metric discovery
- Custom dimensions, metrics, and conversion events
- Data streams and linked Google Ads accounts
Administration (opt-in writes)
- Create, update, and archive custom dimensions
- Create, update, and archive custom metrics
- Create and delete conversion events
- Only used when you explicitly ask for a change
- Never touches your historical analytics data
Best Practices
- List your properties first, then pass the property ID on every report request
- Use the metadata tool to discover the exact dimension and metric names a property supports
- Set explicit date ranges (e.g. last 28 days) for consistent comparisons
- Check metric/dimension compatibility before building complex reports to avoid empty results
- Use realtime reports for "right now" questions and standard reports for trends
- Connect multiple Google accounts if your properties live under different logins — discovery aggregates across all of them