Content published on YouTube, Reddit, and LinkedIn now surfaces in the answers generated by AI. An SE Ranking study of 100,000 queries measures just how much, and shows that Google and ChatGPT do not rely on the same platforms, which means brands need to rethink where and how they publish if they want to be cited by AI.
An active social media presence increases a brand’s chances of appearing in AI answers, but with wide disparities between engines: two-third of Google’ AI Overviews cite at least one social source, compared with fewer than one in ten ChatGPT answers.
The platforms that matter also differ: YouTube for Google, Reddit and LinkedIn for ChatGPT. Social activity therefore does not guarantee a citation; it creates material that the engines can pick up, provided it answers the questions people are actually asking.
What does the SE Ranking study show about social sources in AI answers?
Social media has become a common source for Google’s AI-generated answers, and a marginal one for ChatGPT. That is the main finding of the study published by SE Ranking and SE Visible in early October 2026.
The study analyzed the sources cited across 100,000 keywords spanning 20 industries (100,000 queries for AI Overviews, 50,000 each for AI Mode and ChatGPT). The share of answers containing at least one link to a social or community platform breaks down as follows:
- Google AI Overviews: 67.66% of answers
- Google AI Mode: 32.39%
- ChatGPT: 8.4%
By comparison, in Google’s traditional organic results, 76.73% of queries show at least one social or community result in the top 10. Reddit appears in the top 10 for 46.28% of queries, YouTube and Facebook for around 30%, Quora for 15.79%, and LinkedIn for 6.03%.
In other words, when a Google AI answers a question, it is more likely to point to a Reddit thread, a YouTube video, or a LinkedIn post than to that brand’s website.
Why don’t Google and ChatGPT cite the same platform?
Each engine has its preferred sources, and a strategy that works on one does not necessarily work on the other.
Google favors YouTube. In AI Overviews, YouTube is the most-cited domain in the study, accounting for 16.2% of all citations. No surprise there: the platform belongs to Google, and its content is indexed and transcribed in depth. Reddit leads the social sources in AI Mode.
ChatGPT favors Reddit and LinkedIn. These are the two social platforms ChatGPT cites most, yet each appears in fewer than 1% of answers. Social media carries little weight there compared with reference sites, Wikipedia first among them.
The balance shifts quickly. Search Engine Land reported that Reddit’s share of ChatGPT citations fell by 86.4% in four days in August 2026. Visibility gained on one platform can therefore vanish from an AI answer without the brand having changed a thing.
Industry changes the picture. The averages hide wide variations by topic. LinkedIn is cited in 21.42% of AI Overviews on jobs and careers, versus 3.15% across all topics. YouTube appears in 68.10% of AI Overviews on sports, Facebook in 29.11% of those on food and beverage, and Reddit in 44.85% of those on personal relationships. In real estate, social sources make Google’s top 10 for only 32.70% of queries, compared with more than 90% for jobs.
Branded or generic queries: where does social media weigh most?
Social sources play a bigger role when the user does not name a brand. SE Visible compared two sets of 100,000 queries in Google AI Mode: social sources appear in 39.34% of answers to branded queries, versus 48.58% of answers to generic queries. YouTube, Reddit, and Facebook dominate in both cases, with a different mix.
The takeaway for a brand is simple. When someone searches for it by name, the AI relies first on the brand’s own content and on the media that cover it. When someone looks for a solution to a problem without naming anyone, the AI draws on the conversations, tutorials, and reviews published on social platforms. That is the stage, upstream of brand awareness, where a social presence makes the difference, and it is where most B2B brands are absent.
Why doesn’t posting regularly earn citations?
AI engines cite content that answers specific questions, not brand accounts that post regularly. The study backs this up with a real case study: over twelve months, the software company Planable (part of the SE Ranking group) tried YouTube videos, LinkedIn and Instagram carousels, and even a dedicated subreddit, with no measurable effect on its visibility in AI.
What did work: LinkedIn articles published from employees’ personal profiles rather than the company page. Eight of those twenty-five articles were cited in 422 AI answers, including 268 in Google AI Mode, 121 in AI Overviews, 20 in ChatGPT, and 13 in Perplexity. A single article, a tool comparison, accounted for 78% of the citations.
Three conclusions follow:
- People get cited; brand channels rarely do. Signed, personal expertise is picked up more readily than an institutional message.
- Useful formats win. Comparisons, practical guides, and firsthand accounts: content that answers a purchasing or usage question is what the engines pick up.
- Visibility is uneven and unstable. A piece of content can be cited by three engines and ignored by a fourth; Gemini cited none of the articles tracked.
What does this study change for an AI visibility strategy?
It moves part of the visibility work off the brand’s website and onto the platforms where the questions are being asked. For a B2B company, five practical consequences:
Check your own industry before choosing platforms. The study’s averages do not apply to everyone. An industrial manufacturer, a software vendor, and a medtech company do not share the same mix of cited sources. The first step is to test your market’s queries in ChatGPT, Google AI Mode, Perplexity, and Gemini, and note which platforms and which authors show up.
Let people speak, not just the brand. Executives, technical experts, business leads: their signed contributions on LinkedIn are more likely to be picked up than company-page posts. It is the natural extension of a media relations program, where the spokesperson already carries the credibility of the message.
Produce content that answers an identified question. The starting point is not what the brand wants to say, but what its customers are asking on Reddit, LinkedIn, YouTube, or in specialized forums. Social listening, long confined to reputation management, becomes an editorial planning tool.
Treat YouTube as a search channel. A tutorial or explainer video, a product demo, or an expert interview, properly titled and transcribed, is one of the formats Google picks up most.
Do not abandon the media and third-party sources. Social media does not replace the press, review sites, or Wikipedia in AI citations, particularly for ChatGPT. It adds to them. A coherent AI visibility strategy combines media coverage, owned content, and a personal social presence, then measures what is actually cited, engine by engine.
Key takeaways
Generative AI does not reward social activity as such. It picks up content that answers a question, published where its retrieval systems go looking for answers: YouTube for Google, Reddit and LinkedIn for ChatGPT, with wide variations by industry and over time. For a brand, the challenge is not to publish more, but to publish in the right place, under the right byline, on the right questions.
The useful question is no longer “are we present on social media?” but “on which questions, and by whom, are we being cited?” It is a measurement job before it is a production job.
