google-analytics-ai-insights

Ask questions about your Google Analytics data in plain English. Get instant insights powered by Claude AI.

Installation

npm install google-analytics-ai-insights

Environment Variables

GOOGLE_APPLICATION_CREDENTIALS=./path/to/ga-credentials.json
# OR for production:
GA_CREDENTIALS_JSON='{"type":"service_account",...}'

GA4_PROPERTY_ID=123456789
ANTHROPIC_API_KEY=sk-ant-...

# Optional
TEAM_ACCESS_TOKEN=your-secret-token
PORT=3000
CLAUDE_MODEL=claude-sonnet-4-20250514

Usage

As a standalone server

npm start

As an npm package (e.g. from Laravel)

const {
  createServer,
  queryGA4,
  processQuery,
} = require("google-analytics-ai-insights");

// Option 1: Start the full Express server (includes chat UI)
const { app, server } = await createServer({ port: 3001 });

// Option 2: Start server without the built-in frontend
const { app, server } = await createServer({
  port: 3001,
  serveFrontend: false,
});

// Option 3: Use functions directly (no server needed)
const data = await queryGA4({
  dimensions: ["date"],
  metrics: ["totalUsers", "sessions"],
  startDate: "7daysAgo",
  endDate: "yesterday",
});

const answer = await processQuery("How many users visited last week?");

Laravel Integration

Run the chatbox as a sidecar Node service alongside your Laravel app.

1. Install in your Laravel project:

cd your-laravel-project
npm install google-analytics-ai-insights

2. Create a start script (node-services/ga-chatbox.js):

require("dotenv").config(); // loads your Laravel .env
const { createServer } = require("google-analytics-ai-insights");

createServer({
  port: process.env.GA_CHATBOX_PORT || 3001,
  serveFrontend: false, // Laravel handles the frontend
});

3. Add to your Laravel package.json scripts:

{
  "scripts": {
    "ga-chatbox": "node node-services/ga-chatbox.js"
  }
}

4. Proxy from Laravel to the Node service:

// routes/api.php
Route::any('/ga-chatbox/{path}', function (Request $request, string $path) {
    $response = Http::withHeaders($request->headers->all())
        ->send(
            $request->method(),
            'http://localhost:3001/api/' . $path,
            ['body' => $request->getContent()]
        );

    return response($response->body(), $response->status())
        ->withHeaders($response->headers());
})->where('path', '.*');

5. Run both services (use concurrently, pm2, or Supervisor):

npx concurrently "php artisan serve" "npm run ga-chatbox"

API Endpoints

Method	Path	Description
GET	/api/health	Health check
POST	/api/chat	Natural language analytics queries
POST	/api/query	Direct GA4 API query
GET	/api/schema	Available metrics and dimensions

License

ISC