In this guide
- An MCP server exposes tools that an AI assistant can call. Google maintains a Google Analytics MCP server that reads GA4 data.
- It is read-only, runs locally and needs credentials for your account.
- Use it for quick questions. Check important numbers in GA4 itself.
Asking "which channels brought the most purchases last week?" and getting an answer from your own GA4 data is a small but real time-saver. The Model Context Protocol (MCP) makes it possible: a small program exposes GA4 reports as tools, and an AI assistant such as Claude can call them. Google publishes an official server for it. This page explains what it does, what to be careful about and how it fits with other tools. Details of installation change, so use Google's developer documentation as the source of truth. See also MCP for Google Tag Manager and AI assistant prompts for tracking.
What it is
Google's analytics team maintains an open-source MCP server for Google Analytics. According to Google's documentation and listings, it is read-only: it can run reports, realtime reports and funnel reports, and list accounts and properties, but it cannot change your GA configuration. It runs on your own machine and uses your Google credentials, so the assistant can see only what your account can see.
3 things to remember
Scroll and the points stack up.
An MCP server exposes tools that an AI assistant can call. Google maintains a Google Analytics MCP server that reads GA4 data.
It is read-only, runs locally and needs credentials for your account.
Use it for quick questions. Check important numbers in GA4 itself.
Set-up outline
Enable the Analytics APIs
In a Google Cloud project, enable the Analytics Data API and the Analytics Admin API.
Create credentials
Application default credentials with the read-only Analytics scope.
Install the server
Following Google's instructions, usually with a Python tool such as pipx.
Add it to your assistant's configuration
So the assistant can start it.
Ask a question
Start with something simple, such as sessions by channel for last week.
Good questions to ask
- Which channels brought the most purchases last week, and what was their conversion rate?
- Which landing pages have high sessions but low engagement?
- How did revenue change month on month by device category?
- Show the funnel from view_item to purchase for the last 30 days.
Runs on your domain, anywhere
Your tracking runs on your own domain, so a visit from any country is handled the same way.
Safe use
Credentials
Use the narrowest scope. Keep keys out of chats.
Accuracy
Assistants can misread a report. Cross-check important numbers in GA4.
Privacy
GA4 data should not contain personal data, but check what you send to an assistant provider.
Read-only is a feature
It cannot change your set-up. Keep it that way.
Limits
- It sees only what GA4 collected. Lost events and consent gaps still apply. See GA4 event tracking.
- It does not join GA4 with ad spend or profit.
- It cannot fix tracking problems.
Compared with a first-party dashboard
GA4 answers behaviour questions. For spend, revenue, ROAS and profit by channel and creative, Open Tracked's analytics reads your first-party data, and its assistant Ari answers questions about it in plain words, inside the dashboard, with an allowance that depends on your plan. The two complement each other.
Ask Ari instead of digging through reports
Ari reads your tracking and ad data and answers in plain words.
Frequently asked questions
What is the Google Analytics MCP server?
An open-source server from Google that lets AI assistants read GA4 reports through MCP.
Can it change my GA4 settings?
No. It is read-only.
Where does it run?
On your own machine, using your credentials.
Can I trust its answers?
Verify important numbers in GA4. Assistants can misread data.
Does Open Tracked have its own assistant?
Yes. Ari answers questions about your Open Tracked analytics, and there is an MCP server for your containers and account.
Do I need spreadsheets to read my results?
No. Once tracking is live, the Open Tracked analytics shows spend, revenue, ROAS and profit, and you can ask Ari, the built-in AI assistant, questions about your own numbers in plain words. Ari's allowance depends on your analytics plan.
Is Open Tracked different from Open Track or Open Space Track?
No. Open Tracked is the official name of our server-side tracking platform, and the website is opentracked.io. Many people search for it as Open Track, OpenTrack or Open Space Track. All of them lead to the same product: server-side Google Tag Manager hosting, first-party analytics and a CRM pipeline.
Fix your tracking at the source
Open Tracked hosts your server-side Tag Manager, keeps first-party cookies alive and shows what each ad is really worth.
The Open Tracked team
We build Open Tracked: server-side tracking hosting, analytics and the tools around them, for stores that spend on ads. Questions about this guide, or an idea for the next one?
Comments
Loading comments...
Integrations · 5 min read
