When a buyer asks an AI which software to pick, the answer is built from a handful of sources. On Google, the biggest one is YouTube. This report shows which AI products lean on video, which videos get quoted, and how to make yours easy to quote.
YouTube, with a 20.9% mention share across 3M+ US queries.
Engines quote structure, not popularity.
A growing share of buyers now ask an AI: “what’s the best project management tool for a remote team of 20?” They get one answer, built from a handful of cited sources.
In the sources, on the shortlist. Not in them, invisible.
The biggest source is YouTube. Every major citation study of the last year agrees: it is the most cited domain in Google’s AI answers, and it punches above its weight in Perplexity.
What this report does. It reads four independent datasets (Ahrefs, OtterlyAI, BrightEdge, 5W) from one angle: what a B2B software company should do about it. Every number is sourced.
Humans first. The videos that get cited are the videos people watch. We say so plainly before the playbook.
If you only read one screen, read this one. Everything below is the detail behind it.
YouTube’s mention share across 3M+ US queries. Ahead of Wikipedia, Amazon and every news site.
of AI cited videos have fewer than 1,000 views. Engines quote structure, not view counts.
of cited videos are long form. The sweet spot is 10 to 20 minutes. Shorts get 5.7%.
If you sell software and you’re not producing structured YouTube content, you’re handing the AI answer box to whoever is.Videodeck, from the data below
Three surfaces do almost all of the quoting. One of them is not owned by Google.
A tool with no reason to favor a Google property still leans on structured video.
ChatGPT cites YouTube in about 0.2% of responses. Gemini and Copilot barely cite it at all. They build answers from articles, docs, review sites and Reddit.
The fix is simple: publish a text companion with every video. One production, two citation surfaces.
Two Google surfaces and Perplexity account for almost every YouTube citation in AI search. ChatGPT is the outlier, and the reason every video needs a text twin.
mention share across 3M+ US queries. Two YouTube properties in the top 50 add up to about 40%.
No AI surface has more reach, and its favorite source is a video platform.
It builds answers from text: articles, docs, review sites, Reddit.
AI Overviews, AI Mode and Perplexity together. That is where the volume is.
The video for Google and Perplexity, a companion article for ChatGPT. One production, both surfaces.
The top cited domain in Google AI Mode as well, at 16.6% of responses.
Mayo Clinic, the runner up. YouTube is cited in more than twice as many AI Overviews.
The query types where AI cites YouTube most are the queries software buyers ask.
Where YouTube citations show up, per BrightEdge:
Least likely: abstract concepts, career advice, strategy content.
Translation: AI leans on video hardest when a buyer is evaluating products. “How does X handle SSO,” “X vs Y for agencies,” “is X worth it.” Bottom of funnel, high intent.
Only two social platforms count: Reddit at 46.4% and YouTube at 31.8% of social citations. Together, 78.2%.
You can’t control what Reddit says about you. You fully control your YouTube channel.
Most B2B teams assume YouTube is a popularity contest. The data says the opposite:
A 200 view walkthrough from a 900 subscriber channel gets cited if it answers the question cleanly and is structured so a machine can parse it.
The videos AI quotes are the ones a buyer watches before signing up, and most of them come from channels nobody would call big.
Four in ten videos AI quotes would count as flops on a views dashboard.
Likes are a social signal. Citation is a structural one.
A brand with no channel today can be the cited answer within a quarter.
A 15 query buyer intent audit on text first surfaces found no YouTube at all. The same searches surface 30+ creator videos.
AI citation is not a recommendation engine. It is reference selection. The best structured answer wins, not the biggest channel.Videodeck, reading the OtterlyAI data
Four formats we would build a channel around. Each maps to a buyer query that AI answers with video.
Highest intent on the list. The buyer has a shortlist and is deciding. Build it as a chaptered head to head, one criterion per chapter, so every chapter becomes its own citable answer.
The front door of every software purchase. One tool per chapter, so a single video gets cited for the broad query and for each tool in it. Include your own product in a fair ranking: AI repeats the source’s framing.
Nobody demos your product better than you. Skip the 90 second sizzle reel. A 10 to 20 minute walkthrough with a chapter per feature lets Google cite the exact timestamp.
Heavily cited, almost never made. Pricing pages are vague. A clear “here is what you will actually pay” video fills a vacuum, and content that fills a vacuum gets cited by default.
Real client videos, playing muted. Each one has a presenter from our team and the structure the engines read. See one close to what you need? Ask for a video like it.
The query they win: “X vs Y”, “is X worth it”, “best alternative to X”.
The query they win: “best X for Y”, “top tools for Z in 2026”.
The query they win: “what does X do”, “how does X handle [feature]”, “X setup”.
The query they win: “how to do X”, “X explained”, “how X works”. The most cited category of all.
Recurring, structured content is hard to produce in house at a steady cadence. It is what Videodeck was built for: a production team that plugs into your marketing team and ships citation ready video every month.
See all our workThe engines read structure. People watch the video. So every video we make is written and cut for retention first, then structured so a machine can read it. A citation is what a good video earns on top.
A real presenter, a clear promise in the first twenty seconds, one idea per chapter, screen recordings where words would slow you down.
Buyers trust a person explaining a product more than a page describing it. The same face every week turns a channel into a habit.
Chapters, a proper description and a companion article make the same video readable to an engine. Structure is a layer on top of a good video, not a replacement for one.
AI systems don’t watch your footage. They read the structure around it: title, description, chapters, transcript. People watch it, and they decide whether your channel grows.
OtterlyAI tested which video characteristics correlate with repeated citations. Six things move the needle. The numbers are below, the exact rules right after.
Length, chapters, description, freshness and a text twin. Everything else in the data is noise.
usually across 2 to 5 chapters. Each chapter is its own citable unit.
The strongest correlation of any factor tested. Cited videos average about 334 words.
A smaller but real signal. About half of cited videos use hashtags.
Fresh videos get cited more, especially in fast moving niches. SaaS is one.
Shorts are fine for reach. For AI visibility, build reference videos: explainers, walkthroughs, comparisons.
The highest leverage technique in the data. Google’s AI treats each chapter as its own citable unit, like quoting one H2 from a blog post. 78% of timestamped videos were cited multiple times, typically across 2 to 5 chapters.
YouTube’s rules for chapters to render:
Name chapters the way buyers phrase sub questions (“Pricing breakdown,” “Salesforce integration”). Timestamped citations show up in AI Overviews (73%) and AI Mode (27%) today. Perplexity cites whole videos.
Description length is the strongest signal tested (r = 0.31). The average cited video carries about 334 words of description. Hashtags help a little too (r = 0.20).
Four parts:
Think of it as alt text for the whole video.
Recency correlates with citations (r ≈ 0.3), and SaaS moves fast. A 2024 demo quietly stops being citable for “X in 2026” queries.
Not constant uploads. Maintenance: film again when the product materially changes, update titles and descriptions, add chapters for new features.
ChatGPT barely cites YouTube, but it happily cites a well structured article. Turn every video into a written piece on your site with the video embedded. Google and Perplexity cite the video. ChatGPT cites the article.
Views, likes, subscriber milestones and upload volume all show near zero correlation with citations. They still matter for the channel. They just don’t decide who gets cited.
Starting from zero? This is the sequence we would run. It compounds over months, not overnight.
List the 20 buyer queries you most want to own: “best X for Y,” “X vs your top competitor,” “X pricing.” Run them through Google’s AI Mode and Perplexity and note who gets cited today.
One per format: a comparison, a listicle, a product walkthrough and a pricing explainer. 10 to 20 minutes each, fully chaptered, with metadata grade descriptions.
Publish text companions for all four videos so text first engines can quote you too. Then run your 20 queries again every month and track who gets cited.
The barrier to entry is structure and consistency, not budget or audience size, and most of your competitors haven’t started.
For quoting, checking and copying into your own deck. Correlations are Pearson’s r as published.
| Figure | What it measures | Source |
|---|---|---|
| Which AI products cite YouTube | ||
| 20.9% | YouTube’s mention share in Google AI Overviews across 3M+ US queries, the #1 domain. Two YouTube properties in the top 50 combine to roughly 40%. | Ahrefs Brand Radar, June 2026 |
| 29.5% | Share of Google AI Overviews that cite at least one YouTube video. Next best domain: Mayo Clinic at 12.5%. | BrightEdge AI Catalyst |
| 16.6% | Share of Google AI Mode responses that cite YouTube, where it is also the #1 domain. | BrightEdge AI Catalyst |
| 9.7% | Share of Perplexity answers that cite YouTube. | BrightEdge AI Catalyst |
| 0.2% | Share of ChatGPT answers that cite YouTube. Gemini and Copilot barely cite it at all. | BrightEdge AI Catalyst |
| +100% | Week over week growth in ChatGPT’s YouTube citations, off a small base. | BrightEdge AI Catalyst |
| 38.7% | Share of all YouTube citations across AI platforms that come from Perplexity. AI Overviews drive 36.6%. | OtterlyAI, 2026 |
| 94.9% | Share of all YouTube citations that come from Google AI Overviews, Google AI Mode and Perplexity combined. | OtterlyAI, 2026 |
| 2B+ | Monthly users reached by Google AI Overviews. | Google, 2025 |
| Why software and B2B | ||
| 46.4% / 31.8% | Reddit’s and YouTube’s share of social citations in AI search. Together, 78.2%. | OtterlyAI, 2026 |
| 40.8% | Share of AI cited videos with fewer than 1,000 views. | OtterlyAI, 2026 |
| 36% | Share of AI cited videos with fewer than 15 likes. | OtterlyAI, 2026 |
| 35% | Share of cited channels with under 10,000 subscribers. | OtterlyAI, 2026 |
| r ≈ −0.03 | Correlation between views, likes or subscribers and citation frequency. Effectively zero. | OtterlyAI, 2026 |
| The playbook | ||
| 94% | Share of AI citations that go to long form videos. Shorts get 5.7%, almost all from Google surfaces. | OtterlyAI, 2026 |
| 32.1% / 26.1% | Share of citations going to videos of 10 to 20 minutes, and 5 to 10 minutes. | OtterlyAI, 2026 |
| 78% | Share of timestamped videos cited more than once, typically across 2 to 5 chapters. | OtterlyAI, 2026 |
| 73% / 27% | Where timestamped citations appear: Google AI Overviews and Google AI Mode. Perplexity cites whole videos. | OtterlyAI, 2026 |
| r = 0.31 | Correlation between description length and citation frequency, the strongest factor tested. Cited videos average about 334 words. | OtterlyAI, 2026 |
| r = 0.20 | Correlation between hashtag use and citation frequency. About half of cited videos use hashtags. | OtterlyAI, 2026 |
| r ≈ 0.3 | Correlation between recency and citation frequency, strongest in fast moving niches. | OtterlyAI, 2026 |
| 109 / 0 | Cited sources found in a 15 query buyer intent audit on text first AI surfaces, and how many of them were YouTube videos. | 5W audit, 2026 |
Correlation is not causation, and platform behavior changes quickly. Figures describe repeated citation behavior among already cited videos.
Four independent datasets, read side by side. None of them were commissioned by us.
Domains cited by Google AI Overviews across 3M+ US queries, ranked by mention share of the top 50 sources. The source for YouTube’s #1 position and the 20.9% figure.
Read the study100M+ AI citation instances over 30 days across ChatGPT, Google AI Overviews, AI Mode, Perplexity, Copilot and Gemini, plus metadata analysis of the cited videos. The source for the platform split, the small channel stats and every correlation.
Read the studyYouTube citation rates, query categories and platform comparison across Google AI products, ChatGPT and Perplexity. The source for the 29.5%, 16.6%, 9.7% and 0.2% figures and the content categories.
Read the coverage15 buyer intent queries across five verticals, run on text first AI surfaces. 109 cited sources and not a single YouTube video, while the same searches surface 30+ creator videos. The reason every video here ships with a text companion.
Read the auditHow to read the correlations. Figures use Pearson’s r and describe repeated citation among already cited videos. Correlation is not causation, and platform behavior changes quickly. The AI Mode share in chapter 1 is our own arithmetic: 94.9% for the three surfaces combined, minus the published Perplexity and AI Overviews shares. We will update this report as new data lands.
Videodeck (2026). AI search runs on YouTube: a data guide for B2B software. videodeck.co/ai-search-runs-on-youtubeFour chaptered videos a month with proper metadata is a real production lift. Videodeck handles it for B2B software teams: scripts written for retention, real presenters on a set built for your brand, editing, and the chapters, descriptions and companion articles that make each video easy to quote. You review and hit publish.
hey@videodeck.coCopy
Video as a service solution that plugs into your content marketing and product marketing team so you can accelerate growth through video content.