How do you write a YouTube description that AI will cite?
Updated August 2026
Write the description answer-first: directly answer the video's question in the first one or two sentences, then expand to at least 100 to 200 words, up to 300 or more on a rich video, naming real entities like programs, people, places, and outcomes. Description depth is the single strongest measured factor in whether a video gets cited by AI.
A two-line description barely registers. A substantial, specific one gives the engine the text it needs to understand and quote the video. Put the direct answer and the keywords early, and choose named specifics over adjectives.
In Otterly's analysis of more than 100 million AI citations, description depth was the strongest positive correlate of citation of any variable tested. It is correlational, not proof of cause, but it is the clearest single signal in the data. Separately, a controlled Princeton study found that adding statistics, quotes, and named citations to content can boost how often generative engines surface it, by up to 40%, measured on web text.
- Answer in the first two linesOpen the description by directly answering the video's question. Put the answer and the main keywords up top, before any links or boilerplate.
- Expand to real depthWrite at least 100 to 200 words, more on a rich video. Description depth is the strongest measured signal for whether AI cites a video.
- Name real thingsUse specific names: programs, people, places, products, outcomes. Choose named specifics over adjectives.
- Add facts and sourcesWhere you can, include statistics, quotes, and named sources. Fact-dense content gets surfaced more often.
- Save itUpdate the description in YouTube Studio, with the boilerplate and links below the answer, not above it.