On 23 July 2025, Google released an open-source Model Context Protocol (MCP) server that allows large language models to query Google Analytics data using plain English. The project is published on GitHub and works out of the box with Gemini CLI.
Google Analytics AI integration
Matt Landers, Head of Developer Relations for Google Analytics, introduced the server in a demonstration video. The tool connects language models - including Gemini - to Analytics accounts so users can ask questions such as “How many users yesterday?” without opening the Analytics interface. Follow-up prompts add context, creating a conversational workflow.
The server implements the Model Context Protocol and uses both the Analytics Admin API and the Analytics Data API. Any client that supports the protocol can connect, and full compatibility is confirmed for Gemini CLI.
Main functions
- Retrieve account and property lists
- Run core and real-time metric reports
- Access standard and custom dimensions and metrics
- Generate filter and date-range suggestions
- Provide links to connected Google Ads accounts
- Expose an MCP endpoint for Gemini CLI
Responses are returned in JSON and then converted by the model into plain language.
Setup requirements
Installation takes only a few minutes. Users need Python, pipx, and a Google Cloud project with both Analytics APIs enabled. The service account credentials must have read-only access to the selected property.
Background
Google introduced the Model Context Protocol in 2024 to standardise connections between language models and external services. Gemini CLI followed in May 2025, and the Analytics Data API most recently expanded in April 2025. The new server brings those pieces together for real business queries.
During the demo, the system produced a report of top products for the previous month, first sorted by revenue and then, after a follow-up prompt, by units sold. Landers also ran a scenario with a five-thousand-dollar marketing budget; the model recommended channel allocations that historically generated more than four-hundred-thousand dollars.
.svg)
.svg)
.svg)