Connect Google Analytics 4 to ChatGPT. For your whole team.
One Google sign-in in Loopthink, one connector in ChatGPT, and your team is reporting. The sign-in asks for the read-only scope, so no property settings can be changed from the chat, and the user dimensions that can single out a person, city, age, gender and interests, stay masked until you say otherwise. Every ChatGPT user gets their own login and their own role.
Google ships an Analytics MCP server of its own, but it runs on one machine under one Google login, with Python and the gcloud CLI, and it never reaches Claude in the browser.
Free for 2 servers, no card · Paid plans from €39 · EU-hosted
“Which landing pages lost the most sessions in the last 28 days?”
Three landing pages are down against the 28 days before:
| Landing page | Sessions | Change | Conv. rate |
|---|---|---|---|
| /pricing | 3,184 | −642 | 4.1% |
| /blog/rollout-guide | 2,207 | −418 | 1.2% |
| /integrations | 1,562 | −297 | 2.8% |
Want it by city or age group? Those are masked until an admin enables them.
- read-only
- as m.weber
- user dimensions masked
Google Analytics 4 in a chat client, today.
Today a team that wants Google Analytics in an AI client either adds every colleague as a user on the property, or one person's Google login ends up on a laptop next to a Python install and a gcloud session. The first means Viewer or Analyst rights for people who only need a chart. The second means one login behind every question, no record of who asked what, and nothing that works in the browser. And unlike Search Console, GA4 data can become personal: city, age group, gender and interests of a small audience are a short step from a single person, and nothing in that path knows to hold them back.
Steps checked against the vendor documentation on 17 September 2026: Analytics Data API overview · Analytics Admin API, account summaries · Google's Analytics MCP server · Access and data-restriction management · Developer mode and MCP apps in ChatGPT · Connectors in ChatGPT
One matrix for Google Analytics 4. Roles across, tools down.
| Tool | Marketing | Content | Management |
|---|---|---|---|
| List properties | Marketing may use List properties | Content may use List properties | Management may use List properties |
| Reports (sessions, users, conversions) | Marketing may use Reports (sessions, users, conversions) | Content may use Reports (sessions, users, conversions) | Management may use Reports (sessions, users, conversions) |
| Realtime | Marketing may use Realtime | Content may not use Realtime | Management may not use Realtime |
| Page and landing page reports | Marketing may use Page and landing page reports | Content may use Page and landing page reports | Management may not use Page and landing page reports |
| User and demographic dimensions | Marketing may use User and demographic dimensions | Content may not use User and demographic dimensions | Management may not use User and demographic dimensions |
| Write actions (property, audiences) off by default | Marketing may not use Write actions (property, audiences) | Content may not use Write actions (property, audiences) | Management may not use Write actions (property, audiences) |
Masked for every role: user dimensions (city, region, age, gender and interests) until an admin enables them on the connection.
You set this once. Every ChatGPT user in your team gets exactly this.
Google Analytics 4 runs on the free tier: two MCP servers for one person, no card. The roles, the identity per person and the masking above start at €39 a month, as soon as colleagues join. See the plans
Four checks between ChatGPT and Google Analytics 4.
- 01
Identify
OIDC/SSO against your own IdP, including Microsoft Entra. The call runs as the person who made it, never as a shared service account.
- 02
Authorize
Per-role, per-tool, read or write. Anything not explicitly granted never shows up in the tool list in the first place.
- 03
Mask
Field-level masking and pseudonymization applied on your side, before the result travels. The model never sees the raw field. Composio, AnythingMCP and both provider tunnels have no field-level masking at all.
- 04
Log
Who asked, which tool, which records, which fields were masked. One audit trail across every AI client, exportable for review.
What people actually ask ChatGPT about Google Analytics 4.
- Marketing
“Which channels brought conversions last week, by source and medium?”
- Content
“How did the blog perform this month, sessions and engagement time per article?”
- Management
“Sessions, users and conversions this month versus last month, by country?”
Each question runs as the person who asked it, with that person's role. A Content colleague asking the Marketing question gets the tools their own role allows, not an error, not more.
Four ways to put Google Analytics 4 into ChatGPT.
| Native Google Analytics 4 connector | Build it yourself | Enterprise gateways1 | Loopthink | |
|---|---|---|---|---|
| Per-user identity | Runs locally under the credentials of the person who sets it up | Only if you build it | Yes | Yes, every call runs as the person who asked |
| Tool-level roles | No | Your code | Yes | Yes, write off by default |
| Field masking | No | Your code | Not documented | Yes, before the model sees the data |
| Published price | Free, but local: needs Python, the gcloud CLI and a Cloud project of your own; does not work in Claude in the browser | Your engineers' time | Contact sales | Free, paid from €39 |
| EU company | No | Whatever you are | Mostly US | Yes, Germany |
Native Google Analytics 4 connector: Google's own Analytics MCP server is free, read-only and well made. It runs locally, with Python, pipx and the gcloud CLI, under the Google credentials of whoever sets it up, in a project of their own. It does not work in Claude in the browser, and it has no roles, masking or audit.
1 MintMCP, MCP Manager, Pomerium
Coming from one of them? Composio alternative · MintMCP alternative · MCP Manager alternative · AnythingMCP alternative
Also for
Connect another system to ChatGPT
Before you connect.
- Does every team member need their own Google Analytics 4 account?
- No. One Google sign-in in Loopthink covers the whole team, and it is the only Google account involved. Your people sign in to Loopthink with their own identity, every call is logged under their name, and the seats in Google Analytics 4 stay exactly where they are today.
- Who has to add the connector in ChatGPT?
- One person with the right to add connectors in your ChatGPT workspace, once. After that every team member enables it for themselves. Nobody pastes a key, nobody edits a config file.
- Where does the Google Analytics key live?
- There is none, and that is the point. A Google sign-in grants Loopthink read access, which is stored encrypted in the EU, is never returned by the API and never reaches ChatGPT or the people using the connector. You revoke it any time at myaccount.google.com/permissions, and nobody has to rotate or paste anything.
- Can someone change a property or an audience from the chat?
- No, and not because a switch is off. The sign-in asks for the read-only Analytics scope, so Google itself refuses the write methods, and the connection publishes no write tools either. Both limits would have to fail at once.
- What does the model see?
- The result of the tool call, after masking. Fields you mark as personal are masked before the answer travels to ChatGPT. On top of that, the user dimensions city, region, age, gender and interests are held back for every role until an admin enables them on the connection, so the model gets sessions, channels, pages and countries, not a description of a small audience. The model gets the numbers, not the people.
- Is there an audit trail?
- Yes. Who asked, which tool, which parameters, which fields were masked, when. One log across every client your team uses, exportable for review.
- What does it cost?
- Free for one person and two MCP servers, forever, no card. Starter is €39 per month and adds roles, an identity of their own for every colleague, field masking and a 30-day audit. Team is €99 and adds your own APIs and SQL databases, your own domain and your own identity provider. Prices are on the pricing page, no call needed.
- Google has its own Analytics MCP server. Why Loopthink?
- Google's server is free, read-only and good. It runs locally, though, with Python and gcloud, under a single Google account, and not in Claude in the browser. Loopthink takes the same data and turns it into one connector for the whole team, with roles, masking and audit.
- Do you support Universal Analytics?
- No, only Google Analytics 4. Universal Analytics stopped processing data in 2023, and its API is not part of the connector.
Google Analytics 4 in ChatGPT, with rules.
One key in Loopthink, one login per person, one matrix that decides who may do what.
Free for 2 servers, no card · Paid plans from €39 · EU-hosted