Video and AI citations: why the text around the video gets cited, not the video
Social and video sources are 5.54% of all AI citations across six engines, and the widely shared 94% figure is long-form's share of YouTube citations, not of AI citations overall. What gets a video cited is the text around it, its transcript, description, and chapters, not the footage or the channel's reach.
Video is a small share of AI citations, and the figure being shared to argue otherwise measures something narrower than it claims. Across more than 100 million AI citations tracked over 30 days on six engines, social and video sources together were 5.54% of all citations Tousseyn · OtterlyAI · 2026. The widely repeated claim that 94% of AI citations go to long-form video is a misreading Tousseyn · OtterlyAI · 2026. The 94% is long-form's share of YouTube citations, not of all AI citations. What decides whether a video is cited is the text around it, its transcript, description, and chapters, not the footage or the channel's reach. The sections that follow separate the real finding from the misread and set out what the evidence supports.
The 94% figure describes YouTube, not AI citations as a whole
The 94% figure is real, and it describes YouTube alone. In the study, 94% of YouTube citations went to long-form videos and 5.7% to Shorts Tousseyn · OtterlyAI · 2026. That is a fact about which YouTube videos get cited once YouTube is cited at all. It is not a fact about how often video beats other sources.
The actual share is much smaller. YouTube was 31.8% of the social and video category Tousseyn · OtterlyAI · 2026. Since that category was 5.54% of all citations, YouTube sits near 1.76% of the total. A separate count put social media at about 9% of AI citations in early 2026 Tinuiti · 2026. The share is worth having, and it is not the largest source of citations.
YouTube's citation share rises and falls with the engine
Where video lands depends almost entirely on which engine is answering. On Google AI Overviews, YouTube is the single most-cited domain at about 23.3%, ahead of Wikipedia at 18.4% and Google.com at 16.4% Surfer SEO · 2025. On Gemini and Copilot, YouTube is cited so rarely it is close to absent Tousseyn · OtterlyAI · 2026.
This is the same fragmentation seen across the wider citation data. Of 11,647 domains cited across five engines on commercial questions, 69.6% appeared on only one engine Khallad · SurfacedBy · 2026. Video is a Google and Perplexity opportunity first. Treating it as a uniform channel across every assistant overstates it everywhere except the two surfaces where it works.
AI reads the words around a video, not the video
A video is cited for its text, because text is what the retrieval system can read. Individual images lack the words and authority signals generative search rewards, so visual platforms get skipped while users get the answer in the chat Pinterest · 2026. A video without a readable transcript and description is closer to an image than to a document.
The metadata that correlated with repeat citation was the description, not the picture. Description length was the single video attribute most associated with citation frequency, at a weak-to-moderate correlation Tousseyn · OtterlyAI · 2026. The pattern matches other text surfaces. LinkedIn posts that got cited clustered at 500 to 2,000 words with clear structure Loktionova · Semrush, in collaboration with LinkedIn · 2026. The citable unit is the writing, on YouTube as much as on a blog.
Popularity does not predict citation; reference value does
How often a video is watched has almost no bearing on how often it is cited. Views, likes, and subscriber counts all showed near-zero correlation with citation frequency, around -0.03 Tousseyn · OtterlyAI · 2026. The signals that move human reach do not move machine selection.
Small channels clear the bar routinely. Among cited videos, 40.83% had fewer than 1,000 views Tousseyn · OtterlyAI · 2026. A precise 300-view explainer can be cited over a million-view video that never answers the question directly. The model is selecting a reference, not recommending a creator. Reference value comes from clarity and structure rather than audience size.
Timestamps split one video into several citable units, on Google
Chapters turn a single video into several things a model can cite, and the effect is concentrated on one engine. Among timestamped videos, 78% were cited more than once, usually across two to five chapters Tousseyn · OtterlyAI · 2026. Each chapter behaves like a subheading on a page. The model gets a segment to lift rather than a whole file to summarise.
The benefit is Google-specific. Timestamped citations appeared only in Google AI Overviews and AI Mode, and in none of ChatGPT, Perplexity, Gemini, or Copilot during the study window Tousseyn · OtterlyAI · 2026. Chapter structure is worth the effort where Google is the target surface. It does nothing observable on the engines that do not parse video structure.
The study explains repeat citation, not first pickup
The strongest limit is in the study's own framing, and it changes how the findings should be read. The dataset contained only videos that were already cited Tousseyn · OtterlyAI · 2026. So it explains what makes a cited video get cited again, not how a video is first selected. The correlations are directional, not causal.
What predicts the first pickup sits off the video. In a study of organic discovery, generative-optimisation scores did not predict whether a product surfaced, while referring domains did Sharma · arXiv · 2026. Pages ranking for fan-out sub-queries were 161% more likely to be cited than pages ranking only for the head term Search Engine Land · 2025. Third-party authority and fan-out presence bring a video into range. Its transcript, description, and chapters decide whether it stays. Video earns citations the way text does, by being the clearest available answer and being reachable when the question is asked.
FAQFrequently Asked Questions
Sources
Sources are tiered per our methodology & sources page.
Pinterest: Generative Engine Optimization — A VLM and Agent Framework for Acquisition Growth
Pinterest · 2026
Individual images lack the words and authority signals that generative search rewards, so visual platforms risk being skipped over while users get their answer in the chat. Pinterest's response is to predict what users would search for from each image, group images into theme pages, and link them with authority signals. The live system added 20% organic traffic growth.
Methodology note
First-party engineering paper from Pinterest. Vision-Language Models were fine-tuned to predict likely search queries for each image, aided by agents that mine real-time internet trends. Predicted queries drive collection pages built from multimodal embeddings, with hybrid two-tower nearest-neighbour architectures handling authority-aware interlinking. The system runs in production across billions of images and tens of millions of collections.
The Discovery Gap: How Product Hunt Startups Vanish in LLM Organic Discovery Queries
arXiv · Amit Prakash Sharma · 2026
When users named a product, ChatGPT recognised it 99.4% of the time and Perplexity 94.3%. When they asked discovery questions like best AI tools launched this year, success collapsed to 3.32% and 8.29%. Generative-engine-optimisation scores did not predict discovery. Referring domains, Product Hunt ranking, and Reddit presence did, suggesting traditional SEO foundations carry over to AI visibility.
Methodology note
Independent study of 112 startups randomly drawn from the top 500 on the 2025 Product Hunt leaderboard, tested with 2,240 queries across ChatGPT (gpt-4o-mini) and Perplexity (sonar with web search). Correlations were reported between visibility and signals such as referring domains, Product Hunt rank, GEO scores, and Reddit presence, with p-values.
We Analyzed 127,198 AI Citations. The Five Engines Barely Read the Same Web. (SurfacedBy)
SurfacedBy · Ali Khallad · 2026
SurfacedBy analyzed 127,198 source citations from ChatGPT, Claude, Gemini, Perplexity, and Google AI Mode across roughly 16,400 commercial-intent answers between March and June 2026. Of 11,647 cited domains, 69.6% were cited by only one engine and just 2.7% by all five. Vendor, product, and long-tail pages drew 90.6% of citations; Reddit 1.8% and Wikipedia 0.6%. Gemini averaged 11.0 sources per answer, ChatGPT 3.7.
Methodology note
First-party experiment by SurfacedBy, an AI-visibility tracking vendor with commercial interest, published 27 June 2026 and updated 29 June. About 16,400 answers to real buyer and category questions across five engines; citations counted at the domain level. Authors disclose limits: commercial-query skew, citations are not clicks, engine behavior shifts. Verified by direct fetch.
We Analyzed 89K LinkedIn URLs Cited in AI Search: Here's What Drives Visibility
Semrush, in collaboration with LinkedIn · Margarita Loktionova · 2026
Across 325,000 prompts run against ChatGPT Search, Google AI Mode and Perplexity in January–February 2026, LinkedIn appears in 11% of AI responses on average and ranks second among all cited domains. LinkedIn articles of 500–2,000 words and posts of 50–299 words attract the most citations. Perplexity cites Company Pages in 59% of LinkedIn citations; ChatGPT Search and Google AI Mode cite individual creators 59% of the time.
Methodology note
Vendor research study by Semrush in collaboration with LinkedIn, published March 10, 2026. 325,000 unique prompts across twelve industry categories were sent to ChatGPT Search, Google AI Mode, and Perplexity, yielding 89,000 cited LinkedIn URLs. Each URL was enriched with content type, author signals, engagement, and a semantic similarity score (0.57–0.60). Source verified by direct fetch.
The YouTube Citation Study 2026 (OtterlyAI)
OtterlyAI · Rick Tousseyn · 2026
Across 100M+ AI citations tracked over 30 days across six AI engines, 94 percent of YouTube AI citations went to long-form videos and 5.7 percent to Shorts. Views, likes and subscribers showed near-zero correlation with citation frequency (r approximately -0.03). Description length (r = 0.31) and timestamp presence drove repeat citation: 78 percent of timestamped videos were cited multiple times.
Methodology note
OtterlyAI YouTube Citation Study, published 2 March 2026 by Rick Tousseyn. Direct fetch on otterly.ai confirmed the methodology: 30-day citation tracking across ChatGPT, Google AI Overviews and AI Mode, Perplexity, Microsoft Copilot, and Gemini. Pearson correlation analysis on already-cited videos only — results explain repeat-citation behaviour rather than initial citation eligibility.
In January 2026, social media accounts for around 9% of all AI citations across the platforms tracked. Google AI Overviews cites social media more than four times as often as Gemini. Amazon.com appears zero times in Gemini's citations during the period. Reddit dominates the social share, with YouTube next; patterns differ sharply between engines, including between Google's own Gemini and AI Mode.
Methodology note
Tinuiti's Q1 2026 report uses the Profound platform to track citations across nine categories (apparel, beauty, electronics, food and beverage, home and garden, manufacturing, OTC health, technology, transportation and logistics) and seven AI surfaces (ChatGPT, Perplexity, Google AI Mode, Google AI Overviews, Gemini, Microsoft Copilot, Meta AI). The full report is gated; sample charts and headline figures are public.
AI Overview Fan-Out Rankings Boost Citation Odds by 161% (Surfer SEO study, 10K keywords)
Search Engine Land · 2025
Pages ranking for both the main query and at least one fan-out sub-query collected 51% of AI Overview citations. Pages ranking only for the main query collected just under 20%. Ranking for fan-out queries makes citation 161% more likely than ranking only for the head term. Around 68% of cited pages did not rank in Google's top 10 for any related query.
Methodology note
Search Engine Land coverage, December 2025, of a Surfer SEO analysis of 10,000 keywords and 33,000 fan-out queries extracted with Gemini. Surfer measured the share of AI Overview citations going to pages ranking on the head query, on fan-outs, on both, or on neither, and reported a Spearman correlation of 0.77 between fan-out coverage and citation rate.
Surfer SEO AI Citation Report 2025 (36M AIOs, 46M citations)
Surfer SEO · 2025
Across 36 million Google AI Overviews and 46 million citations between March and August 2025, three domains dominate: YouTube at about 23.3%, Wikipedia at 18.4% and Google.com at 16.4%. Industry mix shifts the picture: NIH leads health at 39%, YouTube and Reddit together carry gaming with 93% and 78% appearance rates, and Shopify takes 17.7% of ecommerce citations.
Methodology note
Surfer's AI Tracker logged AI Overview responses and their citations from March to August 2025, covering 36M Overviews and 46M citations across 57,000-plus URLs. The team broke results into industry segments (finance, health, ecommerce, SEO, gaming, sports, travel) and reported the share of citations earned by the most frequent domains within each category.
About the author Max Ackermann
Max Ackermann is founder and Managing Director of info.link, the product data platform that makes brands visible in AI search and connects every physical product to the web through GS1 Digital Link. He writes about AI search and generative engine optimization (GEO), AI-powered commerce, and how brands can structure product data for ChatGPT, Gemini, Perplexity, and retailer AI assistants like Amazon Rufus. For the past two years he has built the pipelines that put structured product data into AI answers, and run the experiments that test what actually moves AI citations.
Max has 20+ years of experience building digital products and businesses. He previously led McKinsey's Corporate Venture and Design teams across Europe, and as Managing Director of a leading US digital agency he built platforms with Nike, Google, Meta, and Airbnb. He founded the UX Design program at Central Saint Martins College, University of the Arts London, and is a Fellow of the UK's Higher Education Academy. Based in Hamburg, he works closely with GS1 on Digital Link adoption; info.link is headquartered in Hamburg and Berlin and counts GS1 Germany among its investors.
Follow Max on LinkedIn.
