How to set up and use Model Context Protocol (MCP)

What is the Model Context Protocol

Model Context Protocol (MCP)  is a standardized framework designed to enable AI models, particularly large language models (LLMs), to connect with external data, tools, and services such as databases, APIs, and file systems. It acts as a bridge, allowing AI agents to access and utilize these resources effectively.

What is the Model Context Protocol integration

The MCP server acts as a secure bridge between AI agents and your Contentsquare data. 
The Model Context Protocol (MCP) Server Integration allows you to access and analyze your Contentsquare data directly through AI agents like ChatGPT, Claude, Microsoft Copilot, and more. You can leverage natural language queries within your preferred AI-powered tools to get insights from your data.
 

Understanding tool call limits

Your Contentsquare plan includes a tool call allowance specifically for the MCP Server integration.  This determines how many analytics requests you can run through the MCP integration.

Tool call limits are tracked at the account level, not per user or per project. All users connected to your MCP server share the same pool of tool calls.

What is a tool call?

A "Tool Call" means a single execution of a tool provided by the MCP Server.

Tool call limits by plan

How tool calls are counted

Simple queries use 1 tool call:

Example: "What was the bounce rate yesterday for project 1234?"

• Makes 1 call to the analytics tool (ComputeSiteMetrics)
• Counts as: 1 tool call

Complex queries may use multiple calls, but only analytics calls count:

Example: "Compare the bounce rate of my PDPs to my PLPs for the paid social segment"
Makes 3 total calls:

• GetPageGroupsForMapping (configuration - doesn't count)
• SearchSegments (configuration - doesn't count)
• ComputePageComparison (analytics - counts)
• Counts as: 1 tool call

What happens when you reach your tool call limit?

When you reach your monthly query limit, the MCP server will stop processing new requests until:

• Your limit resets at the beginning of the next month or
• You purchase additional capacity (Pro and Enterprise only) for
   

How to install the Model Context Protocol integration

Before you begin

Ensure you have the following before setting up the integration:

• Project access: Permissions to access the Contentsquare projects you want to query via the AI agent
• Compatible AI agent: An MCP-compatible AI tool such as Claude Desktop, ChatGPT (with MCP support), Microsoft Copilot, or other MCP-compatible AI agents
• Tool call volume awareness: Review the tool call limits above to understand your monthly allowance. Pro and Enterprise customers can purchase additional capacity if needed. 

1. Click on 'Analysis setup' in the main menu.
    

2. Click 'Model Context Protocol (MCP)'.

   
    

3. Select your AI agent and follow the set up steps directly on the interface.

   
    

How to use the MCP server integration

Once setup is complete, you can start asking questions to your AI agent in natural language. The agent will automatically use the Contentsquare MCP tools to retrieve and analyze your data.

How to structure your analysis prompts

To request analysis via the MCP server integration, you should specify the Contentsquare project for the AI agent to query. This ensures the prompt references the correct project data.

Here are some examples of common questions you can ask:

Analysis setup

• “What Contentsquare projects do I have access to?”
• “Find a page group in the reference mapping describing the product page.”
• “List all possible site and page metrics .”

Grouping and filtering capabilities:

• "Show me the bounce rate for the last 3 months, grouped by EU countries and by week"
• "Show me the activity rate on a product page for Google Chrome visitors"

Funnel Analysis

• "What's the conversion rate for our checkout funnel this week?"
• "Show me where users are dropping off in the registration flow"

Journey Analysis

• "What are the most common user paths to the product page?"
• "How do users navigate from homepage to purchase?"

Page Performance

• "Compare the bounce rate between our landing pages"
• "Which pages have the highest error rates?"

Impact Analysis

• "Compare conversion rates between mobile and desktop users"
• "What's the revenue impact of our recent homepage changes?"

Error Analysis

• What are the main errors impacting my website?
• What is the impact of the error 'XXX' ?

Metrics and Reporting

• "Show me yesterday's bounce rate and top performing pages"
• "What are the significant changes in our metrics compared to last week?"
   

The MCP server supports the following Contentsquare analysis capabilities:

• Funnels Analysis
• Journey Analysis
• Impact Analysis
• Metrics API
• Page Comparator
• Error Analysis (requires Experiencing Monitoring)
   

FAQs