Google Analytics 4 MCP Server: Connect GA4 to Claude AI

Google Analytics 4 contains everything you need to know about your website. Traffic sources. User behavior. Conversion paths. Attribution data.

Getting answers from GA4? That's another story.

The interface fights you at every turn. Custom reports require clicks through nested menus. Segments need configuration. Comparisons mean manual setup. By the time you find the answer, you've forgotten the question.

The Google Analytics MCP server connects GA4 directly to Claude AI. Ask questions in plain English. Get answers from your actual data.

"What traffic sources drove the most conversions last month?" Just ask.

What Is the Google Analytics MCP Server

The Google Analytics MCP server uses Model Context Protocol to give Claude direct access to your GA4 property.

MCP is Anthropic's open standard for connecting AI to external data sources. Instead of navigating GA4's complex interface, you ask Claude questions. Claude queries your analytics data and returns answers.

For marketers and analysts, this means GA4 insights without the GA4 learning curve.

What Data You Can Access

The Google Analytics MCP server pulls data from GA4's Data API.

Traffic Data:

Sessions, users, and new users
Traffic sources: organic, paid, direct, referral, social
Campaign performance (UTM parameters)
Landing page data

User Behavior:

Page views and unique page views
Average engagement time
Bounce rate and exit rate
Scroll depth and engagement metrics
User flow and path analysis

Conversion Data:

Key events (conversions)
E-commerce transactions and revenue
Goal completions
Conversion rate by source and medium

Attribution:

First-click vs. last-click attribution
Data-driven attribution insights
Assisted conversions
Multi-touch journey data

Audience Insights:

Demographics: age, gender, interests
Geographic data
Device and browser breakdown
New vs. returning users

Use Cases

Track Campaign Performance Across Sources

Marketing dollars spread across channels. Understanding which sources actually drive conversions requires constant monitoring.

Ask Claude: "Compare conversion rates by traffic source for the last 30 days. Which channels have the best ROI?"

Get a cross-channel view without building custom reports in GA4.

Understand User Behavior Patterns

GA4 captures detailed engagement data. But surfacing patterns requires knowing where to look.

"What pages have the highest engagement time but lowest conversion rate? Might be content that engages but doesn't convert."

Let Claude identify opportunities hiding in your behavior data.

Diagnose Traffic Drops

When traffic drops, you need answers fast. GA4's interface makes investigation slow.

"My organic traffic dropped 20% this week. What pages lost the most traffic? Is it a specific country or device type?"

Claude can segment the data and surface what changed.

Measure Content Performance

Content marketing success depends on understanding what resonates.

"Which blog posts drove the most conversions this quarter? What's the average path from blog entry to purchase?"

Turn content metrics into actionable insights.

How to Set Up

Setting up the Google Analytics MCP server takes about 10 minutes.

Step 1: Create a Google Cloud Project

Go to Google Cloud Console and create a new project (or use an existing one). Enable the Google Analytics Data API.

Step 2: Create Service Account

Create a service account in your Google Cloud project. Download the JSON key file.

Step 3: Grant GA4 Access

In your GA4 property, go to Admin > Property Access Management. Add the service account email with Viewer permissions.

Step 4: Note Your Property ID

Find your GA4 property ID in Admin > Property Settings. It's a numeric ID like 123456789.

Step 5: Install and Configure

Install the Google Analytics MCP server and configure it with your service account credentials and property ID.

Step 6: Connect to Claude

Add the server to your Claude configuration and test with a query about your traffic.

Example Queries

Once connected, you can ask Claude:

"What's my total website traffic this month compared to last month?"
"Which landing pages have the highest bounce rate? Are they the same pages with the most traffic?"
"Show me the conversion path for users who made a purchase. What's the typical journey?"
"How does mobile traffic engagement compare to desktop? Are mobile users converting at the same rate?"
"What organic keywords are driving the most traffic to my site?"
"Which referral sources drive the highest-value users based on revenue per session?"

The more specific your question, the more useful the answer.

Common Questions

Does this work with Universal Analytics?

No. Universal Analytics was sunset in 2024. The MCP server connects to GA4 properties only.

Can I see real-time data?

The GA4 Data API provides reported data, which has a slight processing delay. For real-time monitoring, you'd still use GA4's real-time reports directly.

How does this compare to Looker Studio?

Looker Studio creates pre-built dashboards. The MCP server enables ad-hoc questions. Use dashboards for regular reporting, use Claude for investigation and analysis.

Can I compare multiple GA4 properties?

If you configure the MCP server with access to multiple properties, you can ask Claude to compare data across sites.

Is my data secure?

The MCP server runs locally and uses your credentials. Data queries go directly to Google's API. Nothing is stored or sent to third parties.