Which social media platforms does AI cite the most? (September 2026)

Reddit was ChatGPT's most cited social platform in every week of September, though in the first week LinkedIn came within a tenth of a point. Across the month ChatGPT cited Reddit in 5.9% of answers, LinkedIn in 4.9%, and YouTube in 2.2%.

How ChatGPT's top platforms moved from June to September

Reddit fell from 15.3% of ChatGPT answers in the four weeks from June 22 to 6.2% in the four weeks to September 20, most of it in one week in mid-August. YouTube went from 4.6% to 2.5%. LinkedIn rose from 4.5% to 5.3%, close behind Reddit by September.

Google's three AI products don't share one social mix. AI Overviews and AI Mode lean heavily on YouTube; Gemini cites social platforms far less often and favors Reddit.

AI Overviews' favorite social platforms

YouTube was AI Overviews' most cited social platform in every week of September, in 31.2% of answers, ahead of Reddit at 17.7% and Facebook at 13.9%.

How AI Overviews' top platforms moved from June to September

YouTube eased from 35.4% to 31.2% from June to September. Facebook rose from 11.2% to 14.0%, while LinkedIn slipped from 11.8% to 8.3%.

Gemini's favorite social platforms

Gemini cites social platforms much less often than Google's search products. Reddit was its most cited social platform in every week of September, at 5.4% of answers, then YouTube at 2.3% and Facebook at 1.2%.

How Gemini's top platforms moved from June to September

Reddit went from 6.2% to 5.3% and YouTube from 2.5% to 2.2%.

AI Mode's favorite social platforms

YouTube was AI Mode's most cited social platform in every week of September, in 29.5% of answers, then Reddit at 16.5% and Facebook at 10.7%.

How AI Mode's top platforms moved from June to September

Reddit went from 17.7% to 16.2% and Instagram fell from 13.8% to 5.9%. YouTube peaked in August and ended September about level with June on the questions tracked throughout, which is why it is down against August. AI Mode answers carried about half as many links from late August, which pulls every line down, so read the shape rather than the level.

Reddit was Perplexity's most cited social platform in every week of September, but LinkedIn was never more than a point behind. Across the month Reddit and LinkedIn were almost level, at 11.3% and 10.9% of answers.

How Perplexity's top platforms moved from June to September

Reddit rose from 6.4% to 11.0% and LinkedIn eased from 13.1% to 10.9%, leaving the two level.

Microsoft Copilot rarely cites social platforms at all. LinkedIn was its most cited social platform in every week of September, but at only 1.5% of answers, then YouTube at 0.5%. Reddit barely features.

How Copilot's top platforms moved from June to September

LinkedIn slipped from 1.7% to 1.6% and YouTube from 0.7% to 0.4%.

Against August (July 30 to August 28), Reddit lost ground on ChatGPT and AI Mode and slipped on AI Overviews. ChatGPT's fall was the largest, from 10.2% of answers to 5.9%, much of it the mid-August step shown above. Reddit gained on Perplexity, from 9.4% to 11.3%, and on Gemini, from 3.6% to 5.4%.

Against August, YouTube's share of AI Mode answers fell from 37.0% to 29.5% and its share of AI Overviews answers slipped from 33.5% to 31.2%. On ChatGPT the monthly figure edged up, from 1.9% to 2.2%, but that hides a slide within the month: ChatGPT cited YouTube in 2.8% of answers in the first week of September and 1.3% in the last.

The mix also changes with the industry. We grouped the brands behind these questions by industry and found the social platform each engine cites most for each one.

ChatGPT and Perplexity split along business and consumer lines. Both cite LinkedIn most for software, marketing, industrial, and legal questions, and Reddit most for consumer industries such as fashion, beauty, sport, food, and travel. Google's AI Overviews cites YouTube most in almost every industry. The exception is travel, where Facebook comes first, in 30.6% of answers.

The gaps can be large. On marketing and agency questions, Perplexity cites LinkedIn in 23.8% of answers, against 10.9% across all brands. In AI Overviews, TikTok appears in 6.4% of beauty answers and 7.6% of fashion answers, against 1.7% across all brands, and Instagram in 15.8% of food and drink answers, against 7.3%.

So the engine figures above are a starting point. The platforms that matter for your brand depend on what you sell.

Does AI pay attention to LinkedIn, Facebook, and X, or just Reddit and YouTube?

More than you might expect, though it depends which AI you ask. LinkedIn does well almost everywhere. In September it was Perplexity's second most cited website of any kind, used in 10.9% of its answers, and ChatGPT's third. Averaged across ChatGPT, AI Mode, and Perplexity, it turned up in 7.3% of answers.

Facebook and Instagram are mostly a Google story. Facebook was the third most cited website in AI Overviews, used in 13.9% of answers, with Instagram in 7.3%. ChatGPT and Perplexity cited Facebook in under 1%. X hardly features at all, at 0.4% of answers at most.

Why does AI mention my brand but link to other websites?

Because naming a brand and citing its website are two separate things an engine does. When ChatGPT named one of the brands we track in September, it also cited that brand's own website 71% of the time. On Gemini that fell to 43%, the lowest of the six engines, so more often than not Gemini described the brand using someone else's pages.

The reverse holds up much better. When an engine cited a brand's website, it named the brand 67% to 85% of the time. It's worth tracking both, because the gap between them shows whether AI is learning about you from your own site or from other people's.

Do you need a viral post to get cited?

No, and most cited posts are fairly unremarkable. We looked up a random sample of the Reddit threads the engines cited in September, and two in three had fewer than 20 upvotes. The typical one had 9. Threads cited 100 times or more weren't any more popular than the ones cited once.

They were also older and shorter than you'd guess. The median thread was 635 days old, and the median text post was 101 words long.

How do I convince my boss that AI search is worth the effort?

Start with your own numbers. A quick way in is Searchable's free AI visibility report, which shows how often AI names your brand today. In Google Analytics, set up a channel group for AI assistants, covering chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com, and claude.ai, so those visits stop hiding inside referral or direct traffic. AI traffic isn't as large as organic search yet, but it's growing alongside AI chatbot adoption, so track it month to month and show how it's building.

Then test the questions your buyers actually ask. Put 20 to 30 of them to ChatGPT and Google's AI, count how often you come up next to your competitors, and repeat it every month. That gives you an initial number and benchmark to report on, and a gap to close.

• Data: A Searchable analysis of citations by ChatGPT, Perplexity, Google AI Overviews, AI Mode, Gemini, and Copilot on questions tracked by Searchable customers. It describes what these engines cited for these questions, not market share.
• Windows: The heat table covers a 30 day period, August 29 to September 27, 2026, called September in this article. The weekly squares cover the four Monday to Sunday weeks from August 31 to September 27. Gains and losses compare that with the previous 30 days, July 30 to August 28, called August. The line charts cover the weeks from June 22 to September 20, and moves from June to September compare its first four weeks with its last four.
• Measure: The share of an engine's answers that cite at least one page from a platform. A site counts once per answer. A platform is an engine's most cited in a week when it has the highest share of that week's answers.
• Checks on every change: A rise or fall against August is only called when it holds both across all tracked questions and across only the questions asked in both months. Moves from June to September are only described when the questions tracked across the whole quarter move the same way.
• Questions section: Figures on brand mentions come from the same answers. Reddit upvotes, comments, age, and length were read on September 29 for a random sample of cited threads.
• Industries: Each brand is assigned one industry from its website and the topics it tracks. Industry figures are shown for ChatGPT, Perplexity, and AI Overviews separately. A platform is shown as an engine's most cited only when it also comes first with every brand weighted equally and with the brands citing it most set aside; otherwise the two leading platforms are shown as level. The differences named in the text pass the same checks.
• Brand concentration: A platform is only ranked or outlined for an engine when its citations are spread across many brands' questions rather than coming from a few.