NotebookLM is useful because it answers from your sources. That means the quality of the notebook depends more on source preparation than clever prompting.
Start small
Do not upload every file at once. Begin with 15 to 20 high-quality documents: official docs, SOPs, course notes, research reports or customer FAQs. Test whether the notebook can answer common questions before expanding it.
Name files for retrieval
Use names like “2026-video-ads-sop-v1.pdf”. Avoid “final-new-document-2.pdf”. The model may read the content, but humans still need to recognize source, date and version quickly.
Ask for cited answers
A strong prompt: “Based only on the uploaded sources, list five actionable recommendations. Add the source for each recommendation. If the answer is not in the material, say so.”
Use it to find conflicts
Ask: “Where do these documents disagree? List the conflicting claims and their sources.” This is especially useful for old SOPs, mixed departmental language and outdated training material.