# AI Engines Cite Long YouTube Videos, and Views Don’t Track Citations

> What a study of cited YouTube videos found on length, views, descriptions and timestamps, what it can’t prove, and a quick audit to run.

URL: https://llmagnet.com/blog/youtube-ai-citations  
Author: Ben Cohen, Founder & CEO at LLMagnet  
Published: 2026-07-09  
Updated: 2026-11-10  
Category: Guides

AI engines do cite YouTube, mostly long-form videos, and among videos already cited, view count doesn’t track how often one is cited. In OtterlyAI’s vendor study of about 100 million AI citations, YouTube took 31.8% of social citations, and long-form videos took 94% of YouTube’s citations.

The study has one big limit: it only looked at videos already cited by AI engines. It can show what cited videos have in common, not what gets a video cited in the first place. Keep that in mind for everything below.

## Contents

1. How Much YouTube Is Actually Cited
2. Long-Form Dominates, but Longer Isn’t Better
3. Views and Likes Don’t Predict Citations
4. Timestamped Citations Appear Only on Google Search
5. What to Publish
6. A Worked Example: Rewriting a Description
7. Who Should Spend Time on This
8. YouTube Mentions and Link Building
9. A Quick YouTube Audit
10. How to Tell If It Worked
11. The Bottom Line
12. FAQ

## How Much YouTube Is Actually Cited

OtterlyAI collected its citations over 30 days. Social and video platforms made up about 5.54% of the total. YouTube’s 31.8% of that sits behind Reddit at 46.4%. By our arithmetic, YouTube is under 2% of all citations in the sample. It’s a real channel, but a small one.

Where those citations come from matters more. Perplexity and Google’s AI Overviews each produced more than a third of them, and ChatGPT only 4.4%.

![Share of all YouTube citations by AI engine: Perplexity 38.7 percent, Google AI Overviews 36.6 percent, Google AI Mode 19.6 percent, ChatGPT 4.4 percent, Microsoft Copilot 0.5 percent, Gemini 0.2 percent.](https://llmagnet.com/images/blog/charts/youtube-geo-citations-by-engine.png)

*Source: OtterlyAI, updated March 2026. Vendor data.*

Inside each engine the picture shifts again. YouTube was over half of the social citations on Google’s AI surfaces and 22.7% on Perplexity. On ChatGPT it was 9.5%, behind Reddit and LinkedIn.

If your buyers mostly use ChatGPT, YouTube is a lower priority for citations than if they use Perplexity or Google. Check your AI referral sources before deciding how much time to give it.

## Long-Form Dominates, but Longer Isn’t Better

Long-form videos took 94% of YouTube citations and Shorts took 5.7%, mostly on Google’s surfaces. That doesn’t mean you need a 40-minute video. Among cited videos, the largest length bands were 10 to 20 minutes (32.1%), 5 to 10 minutes (26.1%) and over 20 minutes (17.6%), and half ran under 8 minutes. The correlation between duration and citation count was 0.02, which is effectively none.

So make a standard video when citation is the goal, and let the topic set the length.

## Views and Likes Don’t Predict Citations

Among already-cited videos, OtterlyAI found almost no linear relationship between popularity and how often a video was cited:

![Pearson correlation with citation count among already-cited videos: description length 0.31, recency about 0.3, hashtags 0.20, duration 0.02, likes minus 0.02, views minus 0.03, channel subscribers minus 0.03.](https://llmagnet.com/images/blog/charts/youtube-geo-correlations.png)

*Source: OtterlyAI, updated March 2026. Vendor data. Already-cited videos only.*

The signals that did show some relationship were weak but more useful: description length (0.31), recency (about 0.3) and hashtags in the description (0.20). These are correlations, and Pearson’s r only captures straight-line relationships.

The raw counts make the same point. Of the cited videos, 40.83% had fewer than 1,000 views and 36% had fewer than 15 likes. 35% came from channels with fewer than 10,000 subscribers, and the median channel had just 41 videos. A small channel isn’t ruled out. We can’t say how much that helps, because the study has no comparison group of videos that were never cited.

A separate finding seems to conflict with this. [Ahrefs’ December 2025 study](https://ahrefs.com/blog/ai-brand-visibility-correlations) of about 75,000 brands found that YouTube mentions had the strongest correlation with AI visibility on ChatGPT, AI Mode and AI Overviews, at about 0.737 (Spearman). The two studies measure different things. OtterlyAI compares individual videos. Ahrefs compares brands, and counts any appearance of a brand name in a video’s title, transcript or description. It doesn’t limit that to a brand’s own channel, so other people’s videos appear to count.

The Ahrefs sample also only includes established brands (Domain Rating above 40), and a brand that people talk about on YouTube is probably well known everywhere. Ahrefs itself notes that correlation isn’t causation.

## Timestamped Citations Appear Only on Google Search

In OtterlyAI’s data, 31% of cited videos had timestamp signals, and 78% of those timestamped videos were cited repeatedly, often across two to five chapters. All of the timestamped citations appeared on Google: 73% in AI Overviews and 27% in AI Mode. None appeared on ChatGPT, Perplexity, Copilot or Gemini.

So chapters are worth testing if Google is where you want to appear. This data can’t prove they cause citations there, and it shows no sign of them on other engines. They’re cheap to add. [YouTube’s help page](https://support.google.com/youtube/answer/9884579) says you can edit an existing video’s description and add a list of timestamps. The first must be 00:00, you need at least three in ascending order, and each chapter must be at least 10 seconds long. In YouTube Studio, open Content, pick the video, add the list to the description and save. Manual chapters override automatic ones.

## What to Publish

The study found no link with title length (0.02) and doesn’t test title wording or scripts, so this list mixes its findings with our judgment.

- **Write full descriptions.** Description length had the highest correlation reported. Explain what the video covers and who it’s for.
- **Pick narrow topics.** “How to set up abandoned cart emails in Klaviyo” gives an engine a clearer match than “email marketing”.
- **Add chapters named after real questions.** This is for Google’s surfaces, as above.
- **Refresh videos on topics that move.** Recency showed a weak positive correlation.
- **Say your brand name out loud.** Ahrefs counts transcripts, so a spoken mention is part of what it measured.

## A Worked Example: Rewriting a Description

The company, video and details below are made up for illustration. Cited videos in OtterlyAI’s data averaged 334 words of description, so treat this as a start, not a target.

| Before | After |
| --- | --- |
| New video! Everything you need to know about Slotwell. Subscribe for more tips. #productivity | How to connect Slotwell to Google Calendar and stop double bookings. This walkthrough covers the connection, choosing which calendars block your availability, and what happens when two people book the same slot. Applies to Slotwell 4.2 and later. Recorded October 2026. 00:00 What the integration does 00:42 Connecting your calendar 02:15 Which calendars block availability? 04:30 What happens with a double booking? 06:10 How to fix a failed sync |

The first line names the task and the product. The next gives the terms a person would type. The chapter titles are questions someone would ask, and the version and date show whether the advice is still current. OtterlyAI’s advice for fast-moving topics is to update reference videos, with dates in titles, pinned comments and new chapters, rather than to upload more often.

## Who Should Spend Time on This

OtterlyAI says YouTube is cited mainly for demonstrations, walkthroughs, how-to content and comparisons. How that applies to you is our judgment:

| Situation | What to do |
| --- | --- |
| Buyers use Perplexity or Google | Fix descriptions and add chapters on existing videos first |
| Buyers mostly use ChatGPT | Keep YouTube a low priority, and look at Reddit and LinkedIn first |
| B2B software | Make setup walkthroughs, integration demos and comparisons |
| Ecommerce | Make product comparisons and how-to videos for your best sellers |
| No channel and little budget | Get mentioned on channels your buyers already watch, since Ahrefs’ count doesn’t appear to be limited to your own videos |

## YouTube Mentions and Link Building

Ahrefs’ [May 2025 study](https://ahrefs.com/blog/ai-overview-brand-correlation/) of AI Overviews, also 75,000 brands, found Spearman correlations of 0.664 for branded web mentions, 0.527 for branded anchors and 0.218 for backlinks. Its December follow-up called backlink correlations very weak.

Those numbers don’t tell you that a video earns more than a link. A strong brand attracts YouTube mentions, web mentions and search demand all at once, and correlation can’t separate them. Links still matter for Google rankings. A fair reading is that being talked about appears to track AI visibility better than link counts do, so don’t judge AI visibility work by links alone.

## A Quick YouTube Audit

If you already have videos, check these before making new ones:

- Which videos answer one specific question?
- Which have only a line or two of description? Start with those.
- Which long videos lack chapters?
- Do the titles use your audience’s words, or your marketing team’s?
- Is your brand name said in the video, or only shown on screen?

If you have no channel, start with a small set of videos on the questions your best search pages already answer. They need clear structure more than high production value. Then see whether any get cited before you commit to more.

## How to Tell If It Worked

Take 10 videos on similar topics. Rewrite descriptions and add chapters on five, and leave five alone. Run the same 20 questions on Perplexity and Google once a week for four weeks before the change and four weeks after, and log every youtube.com link that’s cited. Count a timestamped link as its own result, since those only appeared on Google. Compare the two groups, and believe only a gap bigger than the week-to-week swing.

Two habits to avoid. Don’t stuff in hashtags because they correlated at 0.20: half of the cited videos had them, so they’re common, and the correlation can’t show they helped. And don’t read Shorts as worthless on Google, where most of their 5.7% share came from.

## The Bottom Line

YouTube is a modest citation source, concentrated on Perplexity and Google, and a low view count doesn’t rule a video out. Full descriptions are the cheapest thing to fix, chapters are worth testing for Google, and brand mentions on other people’s channels may matter more than your own subscriber count. None of these studies shows what causes a first citation, so test on a few videos and measure.

To see where your brand appears in AI answers, try the [free LLMagnet audit](https://ai-visibility.llmagnet.com). LLMagnet measures AI visibility across seven engines, and its agent prepares fixes for you to approve.

## FAQ

### How big is YouTube as an AI citation source?

OtterlyAI’s study of about 100 million citations found YouTube took 31.8% of social citations, behind Reddit at 46.4%. Social platforms were about 5.54% of all citations, so YouTube is under 2% of the total.

### Which AI platforms cite YouTube most?

In OtterlyAI’s data, Perplexity produced 38.7% of YouTube citations, Google AI Overviews 36.6% and Google AI Mode 19.6%. ChatGPT produced 4.4%.

### Do popular videos get cited more?

Among videos already cited, views, likes and subscribers all correlated with citation count at about −0.02 to −0.03, which is effectively zero. The study didn’t look at videos that were never cited.

### What kind of videos get cited?

Long-form videos took 94% of YouTube citations and Shorts 5.7%, but length itself barely correlated with citation count. Description length had the strongest link, at 0.31.

### Do timestamps help videos get cited?

On Google, possibly. In OtterlyAI’s data, all timestamped citations appeared in AI Overviews or AI Mode, and none on ChatGPT, Perplexity, Copilot or Gemini.

### Are YouTube mentions better than backlinks for AI visibility?

Ahrefs found brand mentions on YouTube correlated with AI visibility at about 0.737, against 0.218 for backlinks in an earlier AI Overviews study. These are correlations across brands, so they don’t prove videos outperform links.

### How long should a YouTube description be for AI citations?

Cited videos in OtterlyAI’s data averaged 334 words, and description length had the highest correlation with citation count, at 0.31. That’s a correlation among already-cited videos, so write a full summary with the terms people use rather than aiming for a word count.