Does YouTube Help AI Search Visibility? 0.2% vs 26%

YouTube is a large minority of citations on Google's answer surfaces and close to invisible on some others — and the numbers people quote for the same engine differ by two orders of magnitude because they are dividing by different things.

Before you act on any YouTube citation share, ask one question: percent of what. Otterly.AI puts Gemini at 0.2%; our own much smaller measurement puts Gemini at 26%. Both can be right. One is a share of all YouTube citations, the other is a share of that engine's citations.

Whether publishing video raises your appearance rate is a separate question, and the largest study in this area says explicitly that it did not measure it.

Do AI answers cite YouTube?

Yes, heavily, on Google's answer surfaces. BrightEdge puts YouTube at 29.5% of Google AI Overview citations; our own measurement of one brand put it at 26.6%. The engine spread is enormous, though — in our head-term set, Gemini cited video 26% of the time and ChatGPT 0%. A pooled figure across engines describes neither.

Three claims that keep getting compressed into one

1. AI answers cite YouTube a lot. Well supported. Multiple independent datasets, including ours.
2. *AI answers cite YouTube a lot in general.* False as stated, and every dataset with an engine breakdown shows why.
3. Therefore publishing videos will get your brand into AI answers. Not shown by any of these studies, and the largest one says so itself.

Claim 1 is the finding. Claim 3 is the pitch. Claim 2 is the join, and it is where the arithmetic goes wrong.

Marks: ◐ our own first-party measurement, one brand, one category, one week, and we sell a product in this space. △▲ observational study, no control group, published by a company selling a related product. ○▲ preprint with a statistical result, vendor-affiliated authors. There is no peer-reviewed work on this question that we have been able to find.

What is actually published

BrightEdge, via its AI Catalyst platform, analysing YouTube citation patterns across Google AI Overviews, AI Mode, ChatGPT and Perplexity from May 2024 to September 2025, written up in Search Engine Land in January 2026. Grade △▲ — BrightEdge sells enterprise SEO software, and no sample size, confidence interval or limitation is stated in the write-up.

They also report YouTube cited 200 times more than any other video platform, and note ChatGPT's YouTube citations growing "off a small base."

Otterly.AI, YouTube AI Citation Study 2026: more than 100 million AI citation instances over a 30-day period across six platforms. They identified YouTube videos already cited by those systems, pulled metadata, and ran Pearson correlations. Grade △▲ — Otterly sells an AI visibility tool.

Credit where it is due: this is a more careful study than the genre norm. The authors state that correlation does not imply causation, and they state the limitation that decides the whole question — their dataset contains only videos that were already cited, so the findings are "strongest for explaining repeated citation behavior, not initial eligibility." Almost nobody quoting them repeats that sentence. It is the most important one in the study.

Why 9.7% and 38.7% are not a contradiction

Both are legitimate. They are not comparable, and only the first one tells you where to spend a video budget. Otterly's Gemini figure of 0.2% is compatible with Gemini citing YouTube constantly inside its own answers, if Gemini contributed few citations overall to their 30-day collection. It is not evidence that Gemini ignores video.

We inferred the denominator rather than being told it: their platform figures sum toward 100% and their phrasing is share-of-YouTube-citations language. That is an inference from two clues. If Otterly tells us we read it wrong, we will correct this section and say so here.

Our own numbers, held to the same standard

One brand, one category (AI video clipping tools), July 2026, raw responses retained. We were not studying YouTube; we were counting what got cited and YouTube kept showing up unevenly. All rows use the BrightEdge-style denominator — share of that surface's citations.

Those rows come from four different query sets and must not be averaged. Committing that error while describing it would be an unusually stupid way to write this post.

Our AI Overview figure and BrightEdge's are close — 26.6% against 29.5%, from sample sizes differing by orders of magnitude. That is not replication. Two numbers agreeing once, on different datasets and windows, proves very little.

On the Korean row: the brand we measured had zero Korean-language pages indexed while fourteen other locales — Czech, Greek, Hungarian and Romanian among them — were indexed. On that surface the cited properties were concrete and countable: YouTube 12, Tistory 8, a Korean tech outlet 3. In that market the brand was not losing a ranking contest. It had not entered one. Four queries; direction only.

What all the data agrees on

• On Google's answer surfaces, YouTube is a large minority of citations. BrightEdge 29.5%, Otterly 36.6% of YouTube citations arriving from that surface, us 26.6%. This one is not a single vendor's artefact.
• The engine spread is enormous, and it should drive your budget. In our head-term set, ChatGPT cited vendor product and about pages 66% of the time, review sites 19%, and video 0%. Gemini in the same set cited video 26% and review sites 0%.

One playbook run against both engines moves one and leaves the other flat. The full per-engine breakdown is in Brand Not Showing in AI Search?.

The mechanism, and the measured part of it

There is a plausible story: a YouTube page is not video to a retrieval system. It is a title, a description, a chapter list and a full timed transcript, on a domain with extreme crawl priority, in a format where a specific timestamp can be pointed at as the answer to a specific question.

Otterly's correlations are the best available evidence on this, with their caveat attached:

• Popularity does essentially nothing. Views, likes and subscriber counts land at r ≈ −0.03 to 0.02 — that is noise. If citation were a popularity contest, this is where it would show.
• Text does something. Description length reaches r ≈ 0.31, hashtags r ≈ 0.20. Weak, but on the text axis rather than the audience axis, which is what the mechanism predicts.
• Long-form takes 94% of citations against 5.7% for Shorts. A longer transcript is more surface to match against and more to time-stamp.

All three are correlations within already-cited videos. They describe what distinguishes a repeatedly-cited video from a rarely-cited one. They do not describe what gets a video cited the first time.

What that changes in practice is less about what you publish and more about what you count: if audience signals are noise and text signals are not, a video asset should be measured as a text page with a transcript, not as a channel with subscribers. Views and subscribers are the two metrics a video team already reports, and the two that land at r ≈ 0.

The step nobody measured

Everything written under headings like how to get your videos cited by AI answers the fourth row using evidence from the first three. That may still turn out to be correct. It is not currently shown, and the largest study in the field is the one telling you so.

Our data does not fill that row either: the brand we measured owned no video assets at all. What all of this establishes is how big the slot is. What none of it establishes is the price of getting into it.

How to test it on your own brand

1. Establish a before, as a rate rather than a sighting. Pick 10-15 buyer questions and ask each repeatedly. At a single sample the 95% interval on a brand-detection rate is roughly ±72 percentage points (Schulte et al. 2026, arXiv 2604.07585, grade ○▲). Record per engine, never pooled.
2. Split into treatment and control halves. Publish video targeting only the treatment half. The control half is what tells you whether movement was yours or the engine's.
3. Publish so a null result means something. Full transcript, chapters, long-form, a title that is the question a person would type. If you publish uncaptioned footage and nothing happens, you tested a different hypothesis.
4. Fix the observation window in advance. Nobody has published how long an intervention takes to reach these surfaces. We do not know either. Choosing the window afterwards is how a null becomes a positive.
5. Re-measure identically, including the parts you would rather change.
6. Report the control half, especially if it moved. Roughly 65% of cited sources change between consecutive days in published measurement (Schulte et al. 2026, ○▲).

If step 6 shows your control moving as much as your treatment, you have learned the most valuable thing available here: that you cannot yet attribute anything. The same design problem applies to content updates — see How Often to Update Content for AI Search.

What this post cannot tell you

• Our figures are one brand, one category, one week, across four query sets that are not comparable to each other.
• We did not measure Perplexity or Copilot at all. Those empty cells are facts about us.
• Each AI Overview was measured once. The AI Overview rows are single observations.
• We did not receive either external dataset. We read both write-ups; neither underlying dataset appears to be public.
• The denominator reading above is our inference. We would like to be corrected if it is wrong.
• We cannot tell you that publishing video will change your numbers, and neither can the studies. No before-and-after with a control exists in anything we have read.

Where our own numbers come from

The denominator argument is a claim about instruments, so it is fair to say what ours does. peekr stores every engine response whole, including the citation list attached to it. Counting happens on top of that, which is why the surface table above could be broken out after the fact rather than being locked into one pooled percentage at collection time.

Rates are reported per engine, not merged. Citation counts ship with their denominator — a rate over zero citations is reported as no data rather than as 0%. And the raw responses stay, so when the parser changes the old answers get recounted instead of two series being spliced.

None of that tells you whether to make videos. It tells you percent of what, on your own brand.

[Run the free check on one URL](/en/onboarding/preview) It reads the submitted page for whether an engine can fetch and parse it, then asks one unbranded question about your category and shows the raw answer with the sources attached. One question, one sample — a demonstration of the mechanism, and step 1 above is why that is not yet a baseline.

Method

External figures are attributed inline with publisher, date, method as stated by the publisher, and a conflict-of-interest mark. We read the Search Engine Land write-up of BrightEdge's analysis and Otterly.AI's study page directly. We did not inspect either underlying dataset.

Figures marked ◐ come from our own stored measurements on one brand in the AI video-editing category, July 2026: Google SERPs and AI Overviews via a SERP API across 13 queries with an overview on 12, and engine responses via the OpenAI and Google APIs with retrieval enabled, producing 242 citations across 72 domains in the natural-language set and 154 citations in the head-term set. Raw responses retained. We do not crawl third-party sites, and we did not open, watch, test or review any video or product named or implied here.

If you run the test design above and get a result in either direction, I would like to read it.