Google has launched an open-source Mannequin Context Protocol (MCP) server that permits you to analyze Google Analytics knowledge utilizing giant language fashions like Gemini.
Introduced by Matt Landers, Head of Developer Relations for Google Analytics, the software serves as a bridge between LLMs and analytics knowledge.
As a substitute of navigating conventional report interfaces, you’ll be able to ask questions in plain English and obtain responses immediately.
A Shift From Conventional Stories
The MCP server provides a substitute for digging by way of menus or configuring stories manually. You may kind queries like “What number of customers did I’ve yesterday?” and get the reply you want.
Screenshot from: YouTube.com/GoogleAnalytics, July 2025.In a demo, Landers used the Gemini CLI to retrieve analytics knowledge. The CLI, or Command Line Interface, is a straightforward text-based software you run in a terminal window.
As a substitute of clicking by way of menus or dashboards, you kind out questions or instructions, and the system responds in plain language. It’s like chatting with Gemini, however out of your desktop or laptop computer terminal.
When requested about consumer counts from the day gone by, the system returned the proper complete. It additionally dealt with follow-up questions, exhibiting the way it can refine queries based mostly on context with out requiring further technical setup.
You may watch the total demo within the video beneath:
What You Can Do With It
The server makes use of the Google Analytics Admin API and Knowledge API to assist a variety of capabilities.
In keeping with the challenge documentation, you’ll be able to:
- Retrieve account and property info
- Run core and real-time stories
- Entry customary and customized dimensions and metrics
- Get hyperlinks to related Google Advertisements accounts
- Obtain hints for setting date ranges and filters
To set it up, you’ll want Python, entry to a Google Cloud challenge with particular APIs enabled, and Utility Default Credentials that embrace read-only entry to your Google Analytics account.
Actual-World Use Instances
The server is particularly useful in additional superior eventualities.
Within the demo, Landers requested for a report on top-selling merchandise over the previous month. The system returned outcomes sorted by merchandise income, then re-sorted them by items bought after a follow-up immediate.
Screenshot from: YouTube.com/GoogleAnalytics, July 2025.Later, he entered a hypothetical situation: a $5,000 month-to-month advertising and marketing funds and a objective to extend income.
The system generated a number of stories, which revealed that direct and natural search had pushed over $419,000 in income. It then recommended a plan with particular funds allocations throughout Google Advertisements, paid social, and electronic mail advertising and marketing, every backed by efficiency knowledge.
Screenshot from: YouTube.com/GoogleAnalytics, July 2025.How To Set It Up
You may set up the server from GitHub utilizing a software referred to as pipx, which helps you to run Python-based functions in remoted environments. As soon as put in, you’ll join it to Gemini CLI by including the server to your Gemini settings file.
Setup steps embrace:
- Enabling the mandatory Google APIs in your Cloud challenge
- Configuring Utility Default Credentials with read-only entry to your Google Analytics account
- (Non-obligatory) Setting setting variables to handle credentials extra constantly throughout completely different environments
The server works with any MCP-compatible consumer, however Google highlights full assist for Gemini CLI.
That will help you get began, the documentation consists of pattern prompts for duties like checking property stats, exploring consumer habits, or analyzing efficiency developments.
Wanting Forward
Google says it’s persevering with to develop the challenge and is encouraging suggestions by way of GitHub and Discord.
Whereas it’s nonetheless experimental, the MCP server offers you a hands-on technique to discover what pure language analytics would possibly appear to be sooner or later.
For those who’re on a advertising and marketing workforce, this might assist you to get solutions quicker, with out requiring dashboards or customized stories. And for those who’re a developer, you would possibly discover methods to construct instruments that automate elements of your workflow or make analytics extra accessible to others.
The total setup information, supply code, and updates can be found on the Google Analytics MCP GitHub repository.
Featured Picture: Mijansk786/Shutterstock