Google Analytics MCP: Query GA4 With AI Using Natural Language (2026)
Table of Contents
Every marketing team has the same problem: the data you need is in Google Analytics, but getting it out means navigating a dashboard, picking date ranges, building comparisons, exporting CSVs, and then cleaning the data before you can think about it. By the time you have an answer, you have forgotten the question.
The Google Analytics MCP server changes that. It gives AI assistants like Claude direct, real-time access to your GA4 property — so instead of navigating a dashboard, you ask a question and get an answer.
This guide covers what the Google Analytics MCP server is, how it works, what Google’s official implementation provides, and how WordPress site owners can connect GA4 to Claude in minutes using Easy MCP AI.
What Is the Google Analytics MCP Server?
The Google Analytics MCP server is an implementation of the Model Context Protocol that connects AI assistants directly to your GA4 property via the Google Analytics Data API and Admin API. Instead of exporting data or copying numbers into a chat window, the AI queries your live analytics data on demand.
Google released its own official MCP server for Google Analytics 4, documented at developers.google.com/analytics/devguides/MCP with source code published on GitHub at github.com/googleanalytics/google-analytics-mcp.
The Model Context Protocol itself is an open standard that Anthropic released on November 25, 2024, and donated to the Linux Foundation’s Agentic AI Foundation on December 9, 2025. It defines a shared interface so any MCP-capable AI client — Claude, Cursor, VS Code, n8n, ChatGPT — can talk to any MCP-capable server without custom integration code on either side.
What Google’s Official GA4 MCP Server Can Do
Google’s official implementation uses two APIs to expose GA4 data to AI assistants:
- Google Analytics Data API — for running reports, pivot reports, funnel reports, and real-time reports
- Google Analytics Admin API — for retrieving account structure, property details, and custom dimensions
The server exposes the following tools:
| Tool | What it does |
|---|---|
get_account_summaries | Lists all Google Analytics accounts and properties the authenticated user can access |
get_property_details | Returns metadata about a specific GA4 property |
list_google_ads_links | Returns a list of Google Ads account links for a property |
run_report | Runs a standard GA4 report with dimensions, metrics, date ranges, filters, and ordering |
run_funnel_report | Runs a funnel report via the Data API |
run_realtime_report | Returns live data: active users, current sessions, real-time events |
get_custom_dimensions_and_metrics | Retrieves all custom dimensions and metrics configured on a property |
Important limitation: Google’s official MCP server is read-only. It cannot modify your GA4 configuration, create properties, or edit settings. It is designed strictly for data retrieval and analysis.
Why AI + GA4 Is More Powerful Than the Dashboard
The GA4 interface is built for exploration. It is not built for the kind of multi-step analysis that becomes natural when you have an AI asking questions on your behalf.
With a GA4 MCP connection, you can ask Claude things like:
- “Compare organic traffic from the last 30 days to the same period last year, broken down by landing page.”
- “Which device category has the lowest conversion rate on the checkout page?”
- “Show me real-time active users by city right now.”
- “What are my top 10 events by count this week, and which ones are conversion events?”
Claude queries GA4 directly, formats the response, and can immediately follow up with deeper analysis — all in the same conversation. No CSV export, no pivot table in Sheets, no 5,000-row limit.
One other practical advantage: the MCP server avoids GA4’s standard 5,000-row export ceiling because it queries the Data API directly with pagination, not the GA4 UI export function.
How to Set Up the Google Analytics MCP Server
Option 1 — Google’s Official Server (Developer Setup)
Google’s official GA4 MCP server is a developer-first tool. It requires:
- A Google Cloud Project with the Google Analytics Data API and Admin API enabled
- A service account with appropriate access to your GA4 property
- A service account JSON key file
- Running the server locally or hosting it as a remote server
This is the right path if you are comfortable with Google Cloud IAM, terminal-based setup, and want maximum control. Documentation is at developers.google.com/analytics/devguides/MCP.
Option 2 — Easy MCP AI for WordPress Sites (No-Code Setup)
If your site runs on WordPress, Easy MCP AI is the fastest path. It ships GA4 tools built in — no terminal, no Google Cloud console beyond creating the service account.
Connect GA4 to Claude via Easy MCP AI
Easy MCP AI is a free, open-source WordPress plugin that turns your WordPress site into a remote MCP server. It includes 11 Google Analytics 4 tools alongside 204 total tools covering posts, pages, WooCommerce, SEO, Search Console, SEMrush, DataForSEO, and more.
What Easy MCP AI’s GA4 Tools Can Do
Easy MCP AI exposes these GA4 capabilities through Claude:
- Run custom reports — query clicks, sessions, users, events, engagement, and revenue with dimensions, filters, ordering, and pagination
- Run pivot reports — cross-tabulate any two dimensions with any metric in each cell
- Run real-time reports — get live active user counts and event data
- List properties and data streams — see all GA4 properties and streams your service account can access
- Browse dimensions and metrics — discover valid field names before building queries, so you never get an API error from a typo
- Check conversion events — retrieve the full list of conversion events configured on a property
All credentials are stored AES-256-GCM encrypted with per-provider HKDF-derived keys on your own server. Nothing is sent to any third party until an AI actively calls a tool that requires it. Easy MCP AI never bills you for API usage — it uses your own Google Cloud service account.
Setup: GA4 + Claude via Easy MCP AI
Step 1 — Install Easy MCP AI. Install and activate Easy MCP AI from the WordPress plugin directory. It is free with no paid tiers and no usage limits.
Step 2 — Create a Google Cloud service account. In Google Cloud Console, create a service account, grant it Viewer access to your GA4 property, and download the JSON key file.
Step 3 — Configure GA4 in Easy MCP AI. Go to Easy MCP AI → External Data → Google Analytics 4 in your WordPress admin. Upload the service account JSON and enter your default GA4 property ID.
Step 4 — Connect to Claude. Copy your Easy MCP AI server URL from Easy MCP AI → Dashboard. In Claude, go to Settings → Connectors → Add custom connector, paste the URL, and authorize via OAuth.
Step 5 — Start querying. Open a conversation in Claude and ask about your GA4 data. Claude will call the GA4 tools in real time and return structured answers.
Example Prompts After Connecting
- “What were my top 5 traffic sources by sessions last month?”
- “Show me a pivot table of device category vs. landing page for the last 14 days.”
- “How many active users are on my site right now?”
- “Which pages have the highest bounce rate this week?”
GA4 MCP vs. Exporting to ChatGPT or Claude
It is worth understanding what the MCP approach solves that a simple file upload does not.
| File upload (CSV/export) | GA4 MCP connection | |
|---|---|---|
| Data freshness | Stale at time of export | Live, queried at time of question |
| Row limit | 5,000 rows (GA4 UI) | API-paginated, no hard cap |
| Custom dimensions | Requires manual export setup | Available automatically |
| Follow-up queries | Requires new export | Answered in same conversation |
| Multi-property | One file at a time | Query multiple properties in one session |
| Real-time data | Not available | Available via run_realtime_report |
The practical difference: a file upload is a snapshot. An MCP connection is a live channel.
Key Facts to Know
- Google’s official GA4 MCP server is read-only — it cannot modify GA4 settings or configuration
- It requires a Google Cloud service account with the Analytics Data API and Admin API enabled
- Easy MCP AI’s GA4 integration uses the same service account model, with credentials encrypted AES-256-GCM on your server
- The MCP standard is supported by Claude, Cursor, VS Code, n8n, ChatGPT, and most major AI clients — one service account connection works across all of them
- Easy MCP AI is free and open source, with no paid tiers or usage limits
Conclusion
The Google Analytics MCP server turns GA4 from a destination you visit into a data source your AI assistant can query on demand. Whether you use Google’s official developer implementation or Easy MCP AI’s no-code WordPress integration, the result is the same: instead of navigating dashboards and exporting spreadsheets, you ask questions and get answers.
For WordPress site owners, Easy MCP AI is the fastest path to getting Claude querying your GA4 property — no terminal required, credentials stay on your server, and it works alongside 204 total tools covering your full site workflow.
→ Get Easy MCP AI from the WordPress plugin directory