Connect Google Analytics 4 to Claude Desktop with an MCP Server – A Complete Guide

Perfect for non-coders, marketers, analysts, and founders who want to integrate GA4 insights into Claude Desktop using a Python-powered MCP server — all locally and securely.

What is an MCP Server?

MCP (Model Context Protocol) servers allow AI tools like Claude to interact with real-world data and tools via a standard protocol. Think of it like a bridge between your local data (like GA4 or Shopify) and Claude’s chat interface — so Claude can ask for data and act on it using well-defined tools, all running securely on your machine.

If you want to know more about How Marketers can use  MCP servers, here’s a detailed guide, The End-to-End Model Context Protocol (MCP) Guide for Marketers

What You’ll Achieve

• Pull GA4 reports via Claude (e.g., “Show traffic for the last 7 days”)
• Run everything locally (no cloud or third-party hosting)
• Use Claude as a smart assistant connected to your actual GA4 data

Tools You’ll Need 🛠️

Tool	Why You Need It
Claude Desktop	To access and chat with tools
Python (3.10+)	To run the MCP server locally
Git (optional)	To clone the project repository
GA4 Access	Your credentials to connect to your property

🚀 Need Help With the Setup?

 If the setup feels too technical, our team is here to help.

Don’t worry,  we’ve got your back!

 👉 Click here to get expert assistance now!

Step 1: Install Python + Add to PATH

1. Download Python from: https://www.python.org/downloads
2. Run the installer
3. Important: ✅ Check “Add Python to PATH”
4. Click Install Now

To confirm it’s installed:

Open your command line:

• Windows: Press Win + R, type cmd, and press Enter
• macOS: Press Cmd + Space, type Terminal, and hit Enter
• Linux: Press Ctrl + Alt + T, or open Terminal from your applications menu

Then run this text below:

python --version

If installed correctly, you should see it return something like: Python 3.10.13

( Note: the number next to python might differ based on the version you have downloaded )

Step 2: Download the GA4 MCP Project

Option A: Clone with Git (Recommended)

git clone <https://github.com/smacient/mcp-ga4.git>
cd mcp-ga4

Option B: Download ZIP

• Visit: https://github.com/smacient/mcp-ga4
• Click Code → Download ZIP
• Extract to a folder (e.g., Documents/mcp-ga4)

Step 3: Set Up Python Virtual Environment

1. Open Command Prompt (CMD)
2. Navigate to the project folder:

cd C : \\Users\\YourName\\Documents\\mcp-ga4\\mcp-ga4-main

1. Create a virtual environment by using the line below:

python -m venv .venv

2. Activate it:

.venv\\Scripts\\activate

Your CMD should now show: (.venv) before the prompt

Step 4: Install MCP + Dependencies

Inside the same activated terminal:

pip install uv
uv sync

You should see a lot of dependencies being downloaded, and the sync would have given

Step 5: Get Your GA4 Credentials

Getting your credentials is the only step where you’ll need to log into the Google Cloud Console. Follow this carefully — it’s easier than it sounds!

1. Visit https://console.cloud.google.com and log in with your Google account.
2. Click “Select a project” (top menu) → “New Project”, and give it a name like “Claude GA4 MCP”
3. After it’s created, click “APIs & Services → Library” in the sidebar
4. Search for “Google Analytics Data API”, click it, and press Enable
5. Next, go to “Credentials” in the sidebar
6. Click “Create Credentials → Service Account”, and fill in a name like “Claude GA4 Bot”
7. Skip roles/permissions for now (you can click Done)
8. After creation, click into your new service account → Go to the “Keys” tab → Click “Add Key → Create New Key” → Choose JSON and download the file
9. Rename this downloaded file to credentials.json
10. Move this file into the main folder of your project (where ga4_server.py is located)
11. Then in your Google Analytics property settings, share access with the email address shown in the JSON file (it ends with @<project>.iam.gserviceaccount.com) and give it Viewer or Analyst access

Step 6: Add a .env File

1. Open the mcp-ga4-main folder where you extracted or cloned the project.
2. Go inside the localserver folder.
3. Right-click in the folder → choose New → Text Document → name it exactly .env (delete the .txt extension if it appears).
4. Open this file and paste the following line:

GOOGLE_APPLICATION_CREDENTIALS=C:\\Users\\YourName\\Documents\\mcp-ga4\\mcp-ga4-main\\credentials.json

Tip: On Windows, this line works as-is. You don’t need to double the slashes like you would in Python scripts.

Step 7: Install the Server into Claude

Run this in your terminal (with the virtual env still active):

uv run mcp install localserver/ga4_server.py

This registers the server with Claude Desktop.

Step 8: Configure Claude Desktop

Open Claude → Settings → Developer → Edit Config

1. A folder will open and Add this script below inside claude.json under “mcpServers”:

Step 9 : Ask Claude!

( Pro Tip : After saving the claude.json document, close Claude completely. Make sure you remove any instance of Claude running. You can use the task manager to ensure this. )

Launch Claude Desktop and ask:

Get a GA4 report for the last 7 days.

“Google Analytics 4”: {
  “command”: “uv”,
  “args”: [
    “run”,
    “–with”,
    “mcp[cli]”,
    “mcp”,
    “run”,
    “C:\\Users\\YourName\\Documents\\mcp-ga4\\mcp-ga4-main\\localserver\\ga4_server.py”
  ],
  “env”: {
    “VIRTUAL_ENV”: “C:\\Users\\YourName\\Documents\\mcp-ga4\\mcp-ga4-main\\.venv”,
    “PATH”: “C:\\Users\\YourName\\Documents\\mcp-ga4\\mcp-ga4-main\\.venv\\Scripts;${PATH}”,
    “GOOGLE_APPLICATION_CREDENTIALS”: “C:\\Users\\YourName\\Documents\\mcp-ga4\\mcp-ga4-main\\credentials.json”
  }
}

Claude will:

• Detect the GA4 MCP server
• Ask for permission to use it
• Fetch your data using tools like get_report, compare_report_metrics, etc.

Troubleshooting Tips

❌ No tools visible in Claude? → Double-check your claude.json paths and restart Claude

❌ Env not loaded? → Confirm .env is in the right folder and dotenv.config() is in the script

❌ Permission error from GA4? → Make sure your service account has correct GA4 access in GA settings

That’s It!

You now have a fully running local Python-powered GA4 analytics server talking to Claude. Secure, private, and easy to modify — no cloud hosting required.

Check out our other blogs for more information on MCP servers and various other tools that will end up making your life easy:

• The End-to-End Model Context Protocol (MCP) Guide for Marketers
• Top AI Tool Libraries To Find the Best AI Tools
• Notebook LM: A Guide with Use Cases and Examples

FAQs