Old YouTube Videos Are Now an AI Citation Asset — No New Filming Required

Most brands treat YouTube as a subscriber and watch-time channel. It is also a major source for AI citations, and the quickest opportunity may already be sitting in the archive.

YouTube is now the single most-cited domain inside Google AI Overviews, capturing close to 30% of all AI Overview citations — ahead of Wikipedia and every news publisher tracked. The videos getting cited aren't Shorts or vlogs: 94% of AI video citations go to long-form YouTube content, with instructional/how-to formats and visual demonstrations driving the fastest growth.

This does not require new production. AI systems extract information from existing videos, which makes transcripts and structure the fastest place to start.

What to improve without filming again

• Transcripts. Every video needs an accurate transcript, either uploaded to YouTube directly or published as a companion post linked from the description. This is the single highest-leverage lever — AI cites video roughly three times more often when it has transcript access than when it doesn't.
• Chapter markers. Retrofitting timestamps onto old videos gives AI a structured map of the content, so it can pull and cite a specific segment for a specific question, instead of needing to process twenty minutes of footage to find one answer.
• Titles, descriptions and tags rewritten as questions. "How does X improve Y?" beats a generic branded title, because that's literally the shape of the query AI is trying to match content against.

The principle is the same as strong podcast notes: accurate transcript, clear sections and a useful description. You are making existing work readable and addressable by a machine.

For a brand with past webinars, talks or explainers, this is a focused optimisation task. Audit transcripts, chapters, titles and descriptions first. It turns content you have already paid for into clearer evidence for search engines and AI systems.