Google Analytics MCP Server: Tools, Setup and Security

The Google Analytics MCP server lets an AI app such as Claude Code, Cursor or Gemini read your GA4 accounts, properties and reports. Google publishes it as open-source Python code in its googleanalytics GitHub organization. You run it on your own machine, and it talks to the Google Analytics Admin API and Data API for you.

Checked against: Google Analytics MCP documentation on 28 October 2026. eSEOspace did not install or test this server.

If MCP is new to you, start with what an MCP server is. This profile covers what the server does, how to connect it and what to watch for.

At a glance

What it can do

The GitHub README groups the tools like this:

Account and property information

• get_account_summaries: your Analytics accounts and properties.
• get_property_details: details about one property.
• list_google_ads_links: Google Ads accounts linked to a property.
• list_property_annotations: annotations on a property.
• create_property_annotation: creates an annotation. It needs the --allow-write flag and the analytics.edit scope.

Reports

• run_report: a standard report through the Data API.
• run_funnel_report: a funnel report.
• get_custom_dimensions_and_metrics: custom dimensions and metrics for a property.
• run_realtime_report: a realtime report.

Google's documentation page also lists a run_conversions_report tool, which the README did not list when we checked. The two sources differ, so confirm the tool list in your client after you connect.

Sample prompts from the README include "what are the most popular events in my Google Analytics property in the last 180 days?" and "were most of my users in the last 6 months logged in?"

How to connect it

Google's docs describe these steps:

1. Install Python 3.10 or later and pipx (the docs also mention uv).
2. Enable the Google Analytics Admin API and Data API in a Google Cloud project.
3. Give your user or service account the Viewer role on the property.
4. Set up Application Default Credentials with the read-only scope https://www.googleapis.com/auth/analytics.readonly.
5. Add the server to your MCP client.

For Claude Code, the README gives this command:

claude mcp add analytics-mcp \
  --scope user \
  -e "GOOGLE_APPLICATION_CREDENTIALS=PATH_TO_CREDENTIALS_JSON" \
  -e "GOOGLE_CLOUD_PROJECT=YOUR_PROJECT_ID" \
  -- pipx run analytics-mcp

For clients that use an mcpServers JSON block, the README shows:

{
  "mcpServers": {
    "analytics-mcp": {
      "command": "pipx",
      "args": ["run", "analytics-mcp"],
      "env": {
        "GOOGLE_APPLICATION_CREDENTIALS": "PATH_TO_CREDENTIALS_JSON",
        "GOOGLE_CLOUD_PROJECT": "YOUR_PROJECT_ID"
      }
    }
  }
}

Google's docs name Claude Desktop, Claude Code, Cursor and Antigravity as clients. The docs page notes that Gemini CLI is deprecated, although the README still shows a Gemini setup.

Security notes

• Keep the read-only scope. The server only needs analytics.readonly for reporting. Add analytics.edit only if you truly want the AI to write annotations.
• Leave --allow-write off by default. Without it, the annotation tool is not enabled.
• Your credentials file is the key. The ADC JSON file grants the same access as the Google account behind it. Store it like a password and never commit it to a shared repository.
• Use least privilege. Google's docs ask for the Viewer role on the property, which is enough for reports.
• Data leaves Google. Report rows go to whichever AI model your client uses. Check that this fits your privacy policy.

Our MCP security hub covers the wider risks, and excessive permissions explains why narrow scopes matter.

Alternatives

• Microsoft Clarity MCP server: session recordings and behavior metrics, also local and official.
• Google Ads MCP server: Google's read-only server for paid search data.
• Google Search Console: we found no official Google MCP server as of 28 October 2026, only community-built projects.

Next step

Want GA4, Ads and your CRM in one AI workflow without risky access? See how we approach MCP for marketing teams, browse more MCP server profiles, or talk to us about Claude Code consulting.

Frequently asked questions