Community & UGC

YouTube AEO guidelines: how AI answer engines cite videos, and how to publish for it

YouTube is a major source for Google's AI answers and a minor one almost everywhere else. In the answers SolCrys tracks, Google AI Overviews cites YouTube far more than any other engine and often links to a specific moment in the video, usually early and often at a chapter start. Gemini cites YouTube regularly, while ChatGPT and Claude rarely, if ever, do. The cited videos are most often 5–20 minute long-form, often from small channels with modest views, and their titles echo the buyer's question. For category questions, the tracked brands' own channels weren't cited at all; creators, educators, conference channels, and vendors with category-level videos were. For questions about a brand, engines cite the brand's own channel, its people in other people's interviews and talks, and independent reviews; which one dominates depends on the category. This guide covers how engines reach and read a video, what the cited videos have in common, 10 publishing guidelines, what to do on your own site, and how to measure it, as of Oct. 15, 2026.

Updated 2026-10-15

Questions this guide answers

Do AI search engines cite YouTube videos?
How do I get my YouTube videos cited in Google AI Overviews?
Does ChatGPT cite YouTube?
Do YouTube chapters and captions help AI citations?
Do YouTube Shorts get cited by AI answer engines?
Does embedding a YouTube video on my page help my page get cited?

Direct answer

YouTube is a Google lever. Google AI Overviews cites YouTube far more than any other engine SolCrys tracks, and Gemini cites it regularly. ChatGPT and Claude rarely, if ever, do. Independent research points the same way for Google: an academic study of 61,212 AI Overview citations found youtube.com was the single most-cited hostname (Xu, Iqbal, and Montgomery), and researchers studying health answers describe Google's preference for YouTube as favoring its own platform (Zha and Chang).

Google AI Overviews often doesn't cite a video; it cites a moment in it. Much of the time the link jumps to a timestamp, usually in the first minute or so, and when the video has chapters, often to a chapter start. One long interview with descriptive chapters was cited for many different questions, each at a moment that matched the question. Shorts are cited occasionally, and none in our sample were deep-linked.

Whose video gets cited depends on the question. For category questions ("best X", "X vs Y", "how does X work"), the tracked brands' own channels weren't cited at all. The citations went to independent creators, educators, conference channels, and other vendors' category-level videos. For questions about a brand, engines cite the brand's own channel, its people in other people's interviews and talks, and independent reviews. Which one dominates depends on the category: reviews for consumer brands, interviews and talks for B2B technology companies with a conference presence, and the brand's own channel for small companies with little coverage. The guidelines below follow from that. Answer one question per video, in the buyer's words, early, with chapters. Make category-level videos, put your people on other channels, and put the substance on your own site as text.

Which engines cite YouTube, and how they reach it

YouTube is a Google property, and that shapes everything. Google's own surfaces read YouTube directly; every other engine sees what YouTube exposes to outside crawlers. YouTube's robots.txt has no rules aimed at AI crawlers, but it closes the caption endpoints to all of them, and YouTube's Terms limit automated access to public search engines that follow robots.txt or have written permission. A compliant outside crawler can read a watch page's title, description, and channel, but not fetch the caption file. The table pairs that access with what shows up in the answers SolCrys tracks, as of Oct. 15, 2026.

Engine	How it can reach YouTube	In the answers SolCrys tracks
Google AI Overviews	Google Search's index, which includes YouTube. Google supports key moments set in the YouTube description.	Cites YouTube far more than any other engine, and often links to a specific moment in the video.
Gemini	Google says Gemini uses public information from YouTube to find videos and understand their content.	Cited regularly, as whole videos rather than moments.
Perplexity	Watch pages are crawlable; Perplexity doesn't document how it handles YouTube.	Cited in some independent studies; results depend heavily on setup, and we don't have a reliable read.
ChatGPT	Through the search providers and crawlers it uses; caption files are off-limits to compliant crawlers.	Almost never cited.
Claude	Through its search provider; same robots.txt limits.	Not seen so far, in a smaller sample than the other engines.

Whose videos get cited

We read the metadata of the YouTube videos cited in a 30-day sample of AI answers SolCrys collected, ending Oct. 15, 2026: about 1,700 citations of 740 videos, nearly all of which we could read, from prompt sets in B2B technology, consumer products, and questions about specific people (the local-business sets produced almost none). The patterns on this page hold with and without our own category (AEO). They are traits of the videos that were cited; we didn't sample videos that weren't. Third-party estimates of YouTube's citation share differ several-fold depending on what they count, so we report patterns rather than percentages.

Question type	Whose videos get cited	Typical format
About a B2B technology company with a conference and press presence	Mostly the company's executives and engineers on other people's channels: media interviews, podcasts, conference and user-group talks, booth videos. The company's own channel less often.	Interviews and talks, often long, with the relevant minute linked
About a consumer brand	Independent reviewers and comparison videos. The brand's own channel occasionally.	Reviews, "is it worth it", "X vs Y"
About a small or new company	The company's own channel, often because nothing else exists, even a short clip with few views.	Short product clips
Category questions ("best X", "X vs Y", "how does X work")	Independent creators, other vendors' category-level videos, educators, event and conference channels, and media. The tracked brands' own channels didn't appear.	"Best" lists, explainers, comparisons, tutorials, conference talks

What cited videos have in common

Beyond who made them, a few traits held with and without our own category.

The moment is early. When Google AI Overviews links into a video, the timestamp usually falls in the first minute or so, where the presenter states the topic or the headline answer.
Chapters steer the moment. In videos with chapters, the linked moment tends to land at a chapter start, far more often than chance. About a third of deep-linked videos had no chapters at all, so Google also picks moments from the transcript.
Titles echo the question. "Best" questions cited titles with "best" or "top" far more often than other questions did, and comparison questions mostly cited titles with "vs" or "compare".
Matching is literal. A few citations went to unrelated videos that happened to share a word with the question, such as a pronunciation video for the word "what's" cited for a question that began "What's the best…".
Size isn't a gate. Typical cited videos had under a few thousand views, and many came from channels with fewer than 10,000 subscribers.
Mid-length long-form is the largest group. More than half of cited long-form videos run 5–20 minutes. Shorts are a small minority, and none were deep-linked.
Automatic captions are the norm. Most cited videos rely on YouTube's automatic captions, and the deep-linked ones are no different.
Evergreen videos keep working. Many cited videos were more than a year old.
Engines pick different videos. Google AI Overviews and Gemini cite largely different videos for the same questions.

10 guidelines for YouTube videos

Each guideline ties to the patterns above, a YouTube or Google rule, or published research. For Google's AI surfaces, this is an operating layer on top of SEO, not a replacement: Google's guide to its generative AI features says optimizing for them is "still SEO", and its video guidance applies as written.

1. Plan for Google AI Overviews and Gemini

Set the goal by engine. A YouTube program can change what Google AI Overviews and Gemini cite. It is unlikely to change ChatGPT or Claude, which rarely, if ever, cite YouTube, so those engines need other sources. Check which of your prompts already show YouTube citations on Google before you pick topics; that is where a video has a seat at the table.

2. Make one video answer one question, and title it in the buyer's words

Topical relevance, along with list position, was the biggest driver of being cited first in a 252,000-trial controlled study (What Gets Cited), and cited titles reuse the question's framing: "Best X for Y", "X vs Y", "How to choose X". YouTube's own advice is to pick one or two main words and feature them in both the title and the description. Matching is literal, so pair your brand name with a plain category descriptor ("[Brand] smart thermostat review") rather than relying on a name that collides with common words.

3. Say the answer in the first minute

Google AI Overviews mostly links to moments early in the video, where the presenter states the topic or the answer. Open with the question and a direct answer, then explain. A long warm-up puts the answer later than where those links usually land.

4. Add chapters named after the sub-questions

Google says it will "prioritize key moments set by you, either through structured data or the YouTube description", and in videos with chapters the linked moment tends to land at a chapter start. Write chapter titles as the questions a buyer asks ("Does it work with existing racks?"), not as labels ("Part 3"). YouTube's rules: the first timestamp is 00:00, list at least three in ascending order, and make each chapter at least 10 seconds long. Chapters help Google pick the moment; they aren't a switch that makes a video citable on their own.

5. Write a description that stands on its own

Outside Google, an engine that follows YouTube's robots.txt can read a watch page's title and description but not fetch its caption file. So the description should carry the substance: a first line that restates the topic, a short summary with the key facts, the chapter list, and a link to the page on your site where the detail lives. YouTube allows up to 5,000 characters and asks for a unique description on every video. Update it when facts change.

6. Make category-level videos, not only brand videos

For category questions, the tracked brands' own channels weren't cited at all, while other vendors' channels were, with category-level videos: "best tools" roundups, explainers, and comparisons framed in the buyer's words. A product promo answers "what is [brand]". A video that honestly answers "how do I choose X" or "X vs Y" is the kind that wins category citations, even from a vendor's channel.

7. B2B: put your people on other people's channels

For the B2B technology companies in our sample that speak at conferences and get interviewed, most of what engines cited about them was their executives and engineers in media interviews, podcasts, conference and user-group talks, and booth videos, usually on small channels. Say yes to those slots, and ask hosts to put the speaker's name, company, and topic in the title and description, so the video can be matched to questions about you.

8. Consumer: earn independent reviews and comparisons

For consumer brands, the cited videos about a brand are mostly independent reviews and "X vs Y" comparisons. Make your product easy to review, answer reviewers' questions, and make sure sponsored reviews are disclosed. Your own channel can still be cited for questions about you.

9. Favor 5–20 minute long-form; treat Shorts as a supplement

More than half of cited long-form videos run 5–20 minutes, and that range is also where Google most often links to a moment. Shorts are cited occasionally, mostly by Google AI Overviews, and none in our sample were deep-linked. Cut Shorts from a long video for reach if you like, but don't make them the plan for AI answers.

10. Check the automatic captions for names

Most cited videos rely on YouTube's automatic captions, so uploading your own isn't a prerequisite. But YouTube warns that automatic captions "might misrepresent the spoken content" because of mispronunciations, accents, dialects, or background noise. Review them, make sure your product and company names come through correctly, and upload professional captions where accuracy matters for accessibility or for a technical audience.

On your own site: embeds, watch pages, and text

Embedding a video on a blog post doesn't make that post a video page in Google's eyes, and there's no evidence that the embed itself gets your page cited. Google treats a page as a watch page only when its "main purpose is to show users a single video", and since 2023 it shows video thumbnails only for pages where the video is the main content. An AI Overview that cites the video usually links to youtube.com, not to your page.

What helps your page is the text. Google's AI features guidance lists "making sure that important content is available in textual form" as a best practice. Put the substance of the video on the page as prose (the answer, the steps, the numbers), not just the player. If you build real watch pages for your own videos, Google's standard video markup (VideoObject, plus Clip or SeekToAction for key moments on videos of 30 seconds or more) is how you describe them. That is ordinary video SEO, not an AI-specific schema: Google says there's no special markup for its AI features.

What not to do

YouTube's own policies cover the shortcuts that tempt teams chasing AI citations, and Google's spam policies apply to YouTube results too.

Mass-produced or templated AI videos. YouTube's spam policy bans "using automated tools or AI to churn out high volumes of similar content with minimal changes", and since July 15, 2025, its monetization rules call repetitive, mass-produced uploads "inauthentic content".
Misleading titles, thumbnails, or descriptions. YouTube calls this malicious clickbait. A title written to match a prompt has to match the video too.
Hashtag and keyword stuffing. With more than 60 hashtags, YouTube ignores all of them, and misleading hashtags can get a video removed.
Undisclosed realistic synthetic content. YouTube requires disclosure when a video makes a real person appear to say or do something they didn't, alters footage of a real event or place, or shows a realistic scene that didn't happen. AI help with scripts, titles, thumbnails, or captions doesn't need it.
Instructions aimed at AI in descriptions or transcripts. Google's spam policies now cover "attempting to manipulate generative AI responses in Google Search".

How to measure YouTube's effect on AI answers

Google gives you more here than LinkedIn does. Search Console's platform properties, available to all accounts since July 2026, report how a YouTube channel's videos perform in Google Search, Discover, and Google News, including the queries. YouTube Studio covers views and watch time. Neither tells you whether an AI answer cited the video.

For that, run a fixed set of buyer prompts on each engine repeatedly and record every cited YouTube URL: which video, whether it was a Short, whether the link jumped to a timestamp, and whether the video is yours, a creator's, or a competitor's. Report Google AI Overviews and Gemini separately from the rest, because they behave differently. Answers vary from run to run, as SparkToro's January 2026 study showed for brand recommendations, so judge a change across several runs, not one reading, and when you want to credit a new video, check whether untouched prompts on the same engine moved too.

What may change

YouTube's place in Google's answers depends on Google's choices, which researchers already describe as Google favoring its own platform. A shift there moves every YouTube strategy at once, and engines outside Google change what they retrieve without notice. Treat the engine pattern on this page as current, not permanent, and recheck it against your own prompts. The details here are as of Oct. 15, 2026.

How SolCrys helps

SolCrys runs your buyer prompts on a schedule across ChatGPT, Google AI Overviews, Gemini, Perplexity, and Claude, depending on plan, and records every cited URL, YouTube videos included. The Citations view classifies sources by type and domain and flags which cited pages mention your brand and which name competitors. For each prompt and engine, you can see which videos the answer leaned on and whether they mention you or your competitors.

That is the Measure → Diagnose → Execute → Verify loop applied to video. Measure the citations. Diagnose which prompts you lose and which videos win them. Your team plans and approves the video, the description, and the page that carries its substance; SolCrys never publishes for you. Then the same prompts run again on the same engines, so you can judge across several runs whether the answers moved.

To see which sources ChatGPT cites for 10 of your buyer questions, Start Free: 10 ChatGPT prompts and 1 content audit, free, with email verification and no credit card. ChatGPT rarely cites YouTube, though. Google AI Overviews and Gemini, where most YouTube citations show up, are on the Starter plan and above.

Sources

FAQ

Do AI search engines cite YouTube videos?

Google's do. Google AI Overviews cites YouTube far more than any other engine in the answers SolCrys tracks, and Gemini cites it regularly. ChatGPT and Claude rarely, if ever, cite YouTube, so a video strategy is mainly a strategy for Google's AI surfaces.

How do I get my YouTube videos cited in Google AI Overviews?

Answer one buyer question per video, title it in the buyer's words, say the answer in the first minute, add chapters named after the sub-questions, and write a description that summarizes the video on its own. Google AI Overviews often links to a specific early moment, and in videos with chapters it tends to link to a chapter start.

Do YouTube chapters and captions help AI citations?

Chapters help Google pick the moment to link: in videos with chapters, the linked moment tends to land at a chapter start, and Google says it prioritizes key moments set in the YouTube description. Uploaded captions aren't a prerequisite: most cited videos rely on YouTube's automatic captions, so the job is to check those for misheard product and company names.

Do YouTube Shorts get cited by AI answer engines?

Occasionally, mostly by Google AI Overviews, and in our sample never as a link to a moment. Shorts are a small share of cited videos, so treat them as a reach supplement rather than an AI-answer plan.

Does YouTube's third-party AI training setting affect whether AI engines cite my videos?

No. That setting controls whether YouTube may share your videos with selected AI companies for model training. It is off by default, and YouTube's help page says nothing about search or citations. Whether an engine can find and cite a video depends on the video being public and on what the engine can read from YouTube.

Does embedding a YouTube video on my page help my page get cited?

There's no evidence that the embed itself does. An AI Overview that cites the video usually links to youtube.com, and Google doesn't treat a blog post with an embedded video as a video page. What helps your page is putting the video's substance on it as text: the answer, the steps, and the numbers.

Free ChatGPT visibility check

See where AI answers skip your brand — then fix it, free

Start a free workspace with your domain: 10 buyer-intent prompts through ChatGPT show where you are mentioned, cited, or skipped, and who gets recommended instead. A free content audit in the same workspace hands you the first fix to ship.

Start Free

Free · No credit card · About 5 minutes

Related guides

Community & UGC

LinkedIn AEO Guidelines: How AI Answer Engines Cite LinkedIn, and How to Post for It

Which AI answer engines cite LinkedIn, which posts they pick, the settings that decide whether a post can be cited, and 10 posting guidelines, checked Oct. 15, 2026.

Community & UGC

Comments as AI Citations

User comments under YouTube videos, blog posts, and Reddit threads can appear in AI-cited sources. This guide explains how to participate ethically in comment-layer conversations and what patterns are useful to buyers.

Community & UGC

Reddit AEO Playbook

A 6-step playbook for Reddit as an AEO surface - weekly cadence, thread-authoring patterns, disclosure norms, a 12-week build plan - re-underwritten after ChatGPT stopped showing Reddit citations in August 2026. What Reddit still buys you, and on which engines.

AI Engine Optimization

How to Optimize for Google AI Overviews & AI Mode

How Google's AI Overviews and conversational AI Mode are built from your index, the inheritance signals that decide which pages get cited, and the optimization plays that actually move position.

Citation & Source Influence

How to Build a Source-Layer Strategy for AEO (2026)

A playbook for the three source layers AI engines cite (owned, earned, and community), with evidence, per-engine caveats, sequencing, costs, and an ethics line.

Citation & Source Influence

Schema Markup for AI Search: What Structured Data Actually Does (and Where It Stops)

Schema markup helps AI engines parse and classify your brand, and makes you eligible for richer treatment. It does not, on its own, get you cited or trusted. The honest, complete guide to structured data for AEO: what it does, what it doesn't, which types matter, and how to ship it so it moves the answer.

Measurement