GA4

You set up GA4 MCP Server to fetch and analyze Google Analytics 4 data from language models and other tools. It exposes a range of analysis capabilities, supports flexible date ranges, and outputs results in Japanese for easy interpretation by your workflows.

How to use

Connect your MCP client to the GA4 MCP Server and run the available tools to query GA4 data. Use the main report tool for general analytics, or pick from the 20 analytical tools to tailor insights to your needs. Results are formatted for easy interpretation by your language model, with common metrics like time, percentages, and counts expressed in natural Japanese formats.

How to install

Prerequisites: Node.js 18 or higher, and access to a GA4 property with the required permissions.

1. Clone or obtain the GA4 MCP Server package containing the server code.

2. Install dependencies and build the project.

3. Run the server configuration you need (see MCP configuration examples).

Configuration and runtime

You can run the GA4 MCP Server in two main modes depending on how you authenticate with Google:

• gcloud CLI authentication (recommended, simple): use application-default credentials and specify the GA4 property via environment variables.
• Service account authentication (production): provide the Google service account JSON and the GA4 property ID.

{
  "mcpServers": {
    "ga4": {
      "command": "node",
      "args": ["/path/to/ga4-mcp-server/dist/index.js"],
      "env": {
        "GA4_PROPERTY_ID": "123456789"
      }
    }
  }
}

Security and environment variables

Set up credentials and property access securely. The following environment variables are shown as part of the configuration examples:

• GA4_PROPERTY_ID: the GA4 property you want to query
• GOOGLE_APPLICATION_CREDENTIALS: path to your service account JSON (for production setups)

{
  "mcpServers": {
    "ga4_sa": {
      "command": "node",
      "args": ["/path/to/ga4-mcp-server/dist/index.js"],
      "env": {
        "GOOGLE_APPLICATION_CREDENTIALS": "/path/to/service-account.json",
        "GA4_PROPERTY_ID": "123456789"
      }
    }
  }
}

Notes and limitations

Be aware of GA4 Data API rate limits and data latency. Real-time reporting is available, but standard GA4 data can have delays of 24–48 hours. For precise funnel analyses, consider exporting data to BigQuery.

Tools and capabilities at a glance

GA4 MCP Server exposes a range of basic and analytic tools to query and analyze GA4 data. See the long list below for available functions.

Examples of usage patterns

Basic usage patterns include querying daily traffic summaries, examining top pages, and performing comparisons across periods. You can also drill into hourly traffic, user journeys, and engagement metrics to gain deeper insights.

Available tools

list_accounts

Retrieves a list of GA4 accounts and properties accessible with the configured credentials.

get_property_details

Fetches detailed information about a GA4 property.

run_report

Executes a general-purpose GA4 report query to fetch metrics and dimensions.

run_realtime_report

Queries real-time GA4 data for current activity.

get_metadata

Returns available GA4 dimensions and metrics that can be used in reports.

get_traffic_summary

Provides summary metrics like page views, users, sessions for a period.

get_top_pages

Returns ranking of pages by traffic for a period.

get_traffic_sources

Analyzes where traffic originates from, grouped by channel or source.

get_device_breakdown

Shows device type distribution (mobile, desktop, tablet).

get_geo_breakdown

Analyzes traffic by geographic region.

compare_periods

Compares metrics across two periods (e.g., week-over-week or month-over-month).

get_landing_pages

Analyzes landing page performance.

get_exit_pages

Analyzes exit page performance.

get_user_journey

Analyzes user navigation paths between pages.

get_conversion_funnel

Analyzes conversion funnel steps and drop-offs.

get_hourly_traffic

Breaks down traffic by hour of day.

get_daily_trend

Shows daily trends in metrics over time.

get_new_vs_returning

Compares new users to returning users.

get_engagement_metrics

Returns engagement metrics such as engagement rate and event counts.

get_search_terms

Retrieves site search terms used by visitors.