Skip to main content

Workflow Guide

NotebookLM + WHMCS

Export WHMCS data via MCP and analyze it in NotebookLM

Estimated time: 20 minutes · Workflow: Export + Upload · No config file needed

Overview

NotebookLM does not have native MCP support

Unlike Claude Desktop, OpenClaw, or Cursor, NotebookLM cannot connect to MCP Server directly. This guide covers the two-step workflow: export WHMCS data through a supported MCP client, then upload the results to NotebookLM for analysis. Community MCP servers exist (browser automation) but are experimental and not recommended for production.

Quick Reference

Connection
Upload workflow
MCP Client
Any (Claude, OpenClaw, etc.)
Public URL
Not required
Node.js
Not required
Supported Formats
PDF, Docs, Sheets, .txt
Free Tier
100 notebooks, 50 sources
Key Features
Deep Research, Audio Overviews, Data Tables

Prerequisites

Step 1: Export WHMCS Data via MCP

Open your MCP client (Claude Desktop, OpenClaw, or Cursor) and query WHMCS for the data you want to analyze. Ask for formatted, aggregated output.

Good export prompts

“Export monthly revenue by product group for the last 12 months. Format as a clean report with totals.”
“Summarize all cancelled services this quarter. Group by cancellation reason. Include count and revenue lost per reason. No client names.”
“List ticket categories by volume and average resolution time for the last 6 months.”

Privacy tip: Always request aggregated data without client PII (names, emails, payment methods). Use phrases like “no client names” or “totals only” in your export prompts. See the privacy section for more details.

Step 2: Upload to NotebookLM

  1. Copy the MCP client output to a document (Google Doc, PDF, or .txt file)
  2. Go to notebooklm.google.com
  3. Click Create new notebook
  4. Click Add source and upload your file
  5. Add multiple reports as separate sources for richer cross-analysis

Supported formats

Accepted
PDF, Google Docs, Google Slides, Google Sheets, .docx, .txt, Markdown, URLs, YouTube, audio files (MP3, WAV, M4A)
Not accepted
CSV (rename to .txt or import to Google Sheets), Excel .xlsx (convert to Google Sheets first), ZIP archives

Tip: Upload multiple WHMCS reports as separate sources in the same notebook. For example: revenue data + ticket data + client growth data. NotebookLM can cross-reference all sources in its answers and Data Tables.

Step 3: Analyze with NotebookLM Features

Once your WHMCS data is uploaded, use these NotebookLM features to extract insights:

Deep Research

Combines your uploaded WHMCS data with web research. Ask questions that compare your metrics to industry benchmarks.

“Compare our MRR growth rate to the average for Australian hosting providers. Are we above or below the market?”

Audio Overviews

Generates a podcast-style audio summary of your data. Two AI voices discuss key findings. Shareable with your team.

“Generate an Audio Overview of this week's performance: revenue, new clients, ticket volume, and notable issues.”

Data Tables

Synthesizes your sources into structured tables. Export directly to Google Sheets. Available on Pro and Ultra tiers.

“Create a table with each product group, monthly revenue, client count, average ticket volume, and churn rate.”

Result: You get research-grade analysis that combines your internal WHMCS data with external market context. Audio Overviews let your team consume reports without reading. Data Tables let you export structured data to Google Sheets for dashboards.

Privacy Best Practices

What to upload

  • Aggregated revenue totals by product, date range, or region
  • Ticket volume and resolution time statistics (no client identifiers)
  • Product performance summaries (counts, growth rates, churn rates)
  • Cancellation reason summaries (grouped, no individual records)

What NOT to upload

  • Client names, emails, phone numbers, or addresses
  • Payment method details, credit card numbers, or billing addresses
  • Individual invoice records with client identifiers
  • Support ticket content with client personal information

Alternative: Full privacy with local AI

If you need to analyze raw WHMCS data with client PII, use local AI models (LM Studio / Ollama) instead of NotebookLM. Local AI runs on your hardware and no data leaves your server.

Troubleshooting

“File format not supported”

  • CSV files are not directly supported. Rename .csv to .txt, or import into Google Sheets first.
  • Excel .xlsx files must be converted to Google Sheets before uploading.
  • Each source file must be under 500,000 words or 200MB.

“Source limit reached”

  • Free tier: 50 sources per notebook, 100 notebooks total.
  • Consolidate multiple small reports into one larger document.
  • Upgrade to NotebookLM Plus (included in Google One AI Premium at $19.99/month) for 300 sources.

NotebookLM gives inaccurate answers about my data

  • Check that your uploaded report has clear headers and structure.
  • NotebookLM works best with well-formatted text. Use tables, bullet points, and section headers in your export.
  • Ask your MCP client to format the output specifically for upload: “Format this as a structured report with clear sections and tables.”

Audio Overview quality is poor

  • Audio Overviews work best with narrative-style reports, not raw data tables.
  • Upload a summary document with context and explanations rather than just numbers.
  • Include trend descriptions like “Revenue grew 15% from January to March” instead of just the numbers.

Next Steps