NotebookLM's Major Update: Spreadsheet Analysis, Audio Reports, and Video Output Explained
What Changed—and Why It Matters
NotebookLM has long stood apart from general-purpose AI chatbots by grounding all outputs in sources you upload yourself. That design keeps hallucinations low and makes every answer traceable. The latest update extends that foundation into quantitative data: NotebookLM now accepts Google Sheets directly, opening business financial data to the same structured analysis it previously reserved for documents and PDFs.
The addition of audio and video output formats means the same data can now be consumed as a podcast-style discussion, a short explainer video, or a mind map—whichever format serves the audience.
What NotebookLM Does
NotebookLM is Google's AI tool for structured knowledge management. Unlike general chatbots that draw on broad training data, NotebookLM works only from sources you specify: PDFs, websites, YouTube transcripts, Google Drive files, or now Google Sheets. Because the model stays within those boundaries, outputs come with source citations, and fact-checking is straightforward.
Additional features include mind map generation, quiz creation from uploaded materials, and shareable knowledge bases you can distribute to colleagues.
Looking for AI training and consulting?
Learn about WARP training programs and consulting services in our materials.
The Spreadsheet Update: What's New
Google Sheets Integration
Previously, uploading tabular data required converting spreadsheets to PDF or plain text first. The update makes Google Sheets a first-class source type. The workflow:
- Open NotebookLM and create a new notebook
- Add a source—select Google Drive and choose your spreadsheet
- The sheet appears in the source panel (marked with a green spreadsheet icon)
- Ask questions or set prompts against the data
Sales figures, cost breakdowns, advertising spend, industry comparisons—any structured business data in Google Sheets can now feed directly into analysis.
Automated Analysis from Prompts
Once data is loaded, NotebookLM responds to analytical prompts with specific findings. A prompt like "What patterns appear in sales, costs, and profit across industries?" produces detailed output: which sectors show the highest margins, which have the largest absolute revenue, how advertising spend correlates with returns.
This replaces a manual process—reading through rows, building pivot tables, writing summary text—with a single prompt. The output cites the source data, so users can verify every figure.
Multi-Format Output
The update adds output formats beyond text:
- Audio: NotebookLM generates spoken explanations of the analysis—a podcast-style dialogue between two simulated experts. In tested examples, spreadsheet analysis produced a 21-minute audio report covering industry comparisons, margin differences between large enterprises and specialized firms, and the implications for business strategy.
- Video: Short video reports (around 7 minutes in tested cases) present findings in slide format, narrating key comparisons. An example highlighted a company with $900M+ in revenue but only 13% profit margin versus a smaller specialized firm achieving 69%—making the efficiency gap visible, not just statistical.
- Mind maps: Relationships between data categories visualized as connected nodes.
- Tables: Structured text summaries organized by the categories in the prompt.
Note: Direct bar chart generation from spreadsheet data has limitations in the current version. Text and table outputs are fully functional; complex visual charts are an area for future improvement.
Practical Workflow
Preparing Your Data
The analysis quality depends on data quality. Spreadsheets should have clear column headers, consistent data types, and logical organization. Before uploading, confirm that the data you want analyzed is well-structured.
Prompt Engineering
The prompts you write determine what NotebookLM extracts. Specific analytical questions produce specific outputs. Examples of effective prompts:
- "Compare advertising spend as a percentage of revenue across each industry in this dataset"
- "Identify the three industries with the highest profit margins and explain what they have in common"
- "What are the outliers in this dataset, and what might explain them?"
Users who also work with ChatGPT report using it to generate analytical frameworks—which industries to compare, which metrics matter—then feeding that structure into NotebookLM as a prompt.
Reviewing and Acting on Output
NotebookLM shows source citations alongside each finding. Before using analysis in presentations or strategy documents, verify the cited figures against the original spreadsheet. The citation mechanism makes this faster than it would be with a general-purpose AI.
Organizational Applications
Consistent analysis from shared data: When different teams analyze the same dataset, they often reach different conclusions based on different framings. NotebookLM produces consistent outputs from the same sources, which reduces conflicting narratives in cross-functional meetings.
Accessible reporting for non-analysts: The audio and video formats make data analysis accessible to colleagues who don't read detailed reports. A 7-minute video or a podcast-style discussion covers the same ground as a multi-page document, in a format that reaches different audiences.
Knowledge bases for teams: NotebookLM notebooks can be shared. A finance team that builds an analysis notebook can share it with product, sales, or leadership—giving everyone access to the same grounded Q&A interface without distributing raw spreadsheets.
Limitations to Know
- Bar chart and complex visualization output is not yet fully supported for spreadsheet data—expect text and table outputs as the primary formats
- The model works only from uploaded sources; it won't supplement your data with external information
- Very large spreadsheets may require selective upload—the model performs better on focused datasets than on warehouse-scale exports
Summary
NotebookLM's spreadsheet integration closes a significant gap for business users. The combination of Google Sheets support, automatic insight extraction, source-cited outputs, and multi-format delivery (text, audio, video, mind map) creates a complete workflow from raw business data to shareable analysis.
The practical result: analysis that previously required a data specialist and several hours of work can now be initiated by anyone with access to the spreadsheet and completed in minutes. The quality depends on prompt quality, but the production bottleneck is effectively removed.
Reference: https://www.youtube.com/watch?v=n6Z3hdLgWk0
TIMEWELL AI Consulting
TIMEWELL supports business transformation in the AI agent era.
Our Services
- ZEROCK: High-security AI agent running on domestic servers
- TIMEWELL Base: AI-native event management platform
- WARP: AI talent development program