ds@mcpanalytics.aiEmail your data and your question and get an answer about your own data, with the method shown.Email your data and question. Get an answer.

Get Started with
MCP Analytics

Connect Claude to professional statistical analysis tools in 3 simple steps. Sign in with Google, GitHub, or email, no verification wait.

1

Install MCP Analytics - Choose Your Method

One-click installation or manual configuration - you choose!

πŸ“¦ Prerequisites

Before installing MCP Analytics, ensure you have:

  • Node.js and npm installed - Required for the npx command
  • Node.js 18 or newer (npx ships with npm 5.2+)

Download Node.js β†’ or check if installed: node --version

πŸš€ Option A: npx with your API key (Recommended)

Get your API key, then add the config below

  1. Sign up at account.mcpanalytics.ai
  2. Click Reveal API Key and copy the key. It starts with mcp_
  3. Paste it into the config below and restart Claude Desktop
  4. See the install page for other options

Platform-Specific Installers

  • πŸͺŸ Windows: PowerShell installer
  • 🍎 macOS: Shell script installer
  • 🐧 Linux: Bash installer
  • πŸ“ VS Code: Marketplace extension

Visit install page for all options

βš™οΈ Option B: Manual Configuration

For advanced users - add to Claude Desktop config:

Location: Settings β†’ Developer β†’ Edit Config

claude_desktop_config.json
// Replace mcp_your_key_here with your API key { "mcpServers": { "mcpanalytics": { "command": "npx", "args": ["-y", "@mcp-analytics/mcp-analytics"], "env": { "MCP_ANALYTICS_API_KEY": "mcp_your_key_here" } } } }
βœ“ Configuration ready - save and restart Claude Desktop
2

Connect with your API key

Your API key signs Claude Desktop in. No browser login is needed.

What happens:

When you restart Claude Desktop, it starts the server with the API key in your config:

  • The server starts - npx runs it with your API key
  • MCP connection established - tools available after the restart

Check connection status:

Type /mcp in Claude to see server status and available tools

Claude Desktop
⏺ Checking MCP server status...
mcpanalytics - about
(MCP)(topic: "platform")
{
"company": "PeopleDrivenAI",
"description": "Enterprise AI-powered statistical analysis platform",
... +48 lines (ctrl+r to expand)
}
mcpanalytics - discover.tools
(MCP)(query: "list available tools", limit: 10)
[
"linear_regression",
"ridge_regression",
"lasso_regression",
"arima",
"correlation",
"t_test",
"anova",
"chi_square",
"pca",
"time_series_forecast"
]
βœ“ Comprehensive statistical toolkit available via MCP
3

Run Your First Analysis

Start analyzing data with natural language requests

How it works:

Once you are connected, you get a 30-day demo on your own data, no card required. Here are the ways to get data into your analysis:

πŸ“¦ Option 1: Demo Datasets (Quickest)

3 pre-loaded datasets are available immediately on signup:

  • Advertising Spend - Marketing spend vs. sales across channels
  • Survey Responses - Categorical survey data with demographics
  • Retail Demand - Daily sales for time series forecasting

Ask Claude to list your datasets or just say "run a demo analysis" to get started.

πŸ”— Option 2: Connect Live Data Sources

Pull data directly from platforms you already use:

  • Google Analytics 4 - Traffic, engagement, and conversion data
  • Google Search Console - Search performance, clicks, and rankings

Use connectors_list to see available connections. No file download needed.

πŸ“€ Option 3: Upload Your Own Data

Secure upload:

  • Generate upload token with datasets.upload
  • Upload CSV, TSV, or JSON via cURL
  • Access with uuid://dataset_id:your_secret

Example walkthrough:

Let's run an analysis using a demo dataset and see what you get back.

β–Ά You
I want to see how my advertising spend relates to sales. Can you run a regression on my demo data?
⚬ Claude
Sure! I'll use your Advertising Spend demo dataset to run a linear regression and see which channels drive the most sales.
⏺ Claude
mcpanalytics - path_resolver
(MCP)(objective: "How does ad spend relate to sales?", dataset_ref: "uuid://<id from datasets_list>")
picked path: tool_run
mcpanalytics - path_tool_run
(MCP)(
objective: "How does ad spend relate to sales?",
dataset_ref: "uuid://<id from datasets_list>",
tool_names: ["linear_regression"]
)
βœ… SUCCESS: path_tool_run
πŸ“Š Interactive Visual Report Available!
https://account.mcpanalytics.ai/report/rpt_AdSpend_Regression
Model Results:
β€’ RΒ² Score: 0.897 (89.7% variance explained)
β€’ Top Channel: TikTok ($4.20 return per $1 spent)
β€’ Significant Predictors: 3 of 3
... +300 lines available in report
⚬ Claude
The analysis is complete. Here is your interactive report: Key findings from the advertising spend analysis:
β€’ Model Quality: RΒ² = 0.897 (89.7% variance explained)
β€’ TikTok: Highest ROI at $4.20 return per $1 spent
β€’ Facebook: Significant but diminishing returns above $50K
β€’ Google Ads: Steady baseline contributor to revenue

What you can do with your report:
β€’ Card viewer - Browse results card by card with an AI insights panel
β€’ PDF export - One-click download with charts on each page
β€’ Share - Send a token link to colleagues (no login required to view)

You're ready to go!

Your 30-day demo works across every analytical tool. Explore what MCP Analytics can do: