LinkedIn Is the #2 Most-Cited Domain in AI Search. Your Company Page Isn't Why.

Every B2B social report measures the company page. The AI citation data says the company page is the minority of your LinkedIn visibility. Meltwater analyzed 9.5 million AI citations across six platforms and found 75% of LinkedIn citations pointed at individual member profiles rather than company pages. Semrush, working from a different corpus, found a LinkedIn URL in 11% of AI responses across 325,000 prompts — second only to Reddit. Both put LinkedIn near the top of the cited-domain list. Neither puts your brand page there.

Key takeaways

The headline "LinkedIn is the #2 most-cited source in AI" is circulating without the number attached to it, which is a problem, because the two studies behind it report figures an order of magnitude apart and both are correct.

Semrush is counting responses that contain at least one LinkedIn link. Meltwater is counting citations, of which there are millions, spread across a very long tail of domains. A domain can be second on the entire web and still hold half a percent, and the gap between 11% and 0.53% is arithmetic rather than disagreement.

That distinction matters for what you do next. The 11% figure is the one relevant to a visibility decision — it says that in a professional-topic prompt set, roughly one answer in nine already has a LinkedIn page inside it. The 0.53% is the one to quote if someone tells you LinkedIn is where AI search happens. It is a top domain in a field with no dominant domains, the same shape our review of where AI citations come from found across the board.

The finding with an actual organizational consequence is the account-type breakdown, and it is where the two studies diverge in a way worth reading carefully.

Meltwater's 75/25 is the number being repeated. Semrush's per-engine breakdown is the more useful one, because it shows the tilt is real on the two highest-volume surfaces and inverts on Perplexity, where company pages take the majority. A brand whose buyers live in Perplexity would draw the opposite conclusion from the headline.

What survives both: on the engines most people are actually tracking, the majority of your LinkedIn citations are not coming from the account your social team runs. Meltwater's follower data sharpens it — 51% of citations came from members with fewer than 10,000 followers. This is not an influencer finding. It is closer to the opposite: engines are citing ordinary practitioners who write clearly about their own subject.

That has a governance edge nobody in the vendor write-ups mentions. Citations earned on a personal profile belong to a person, and people leave. Company page citations are an asset on the balance sheet; profile citations are a relationship. Both are worth having, and only one of them is durable.

The engagement data is the part most likely to change behavior, because it breaks the proxy metric everyone uses.

Semrush reports the median cited LinkedIn post carries 15 to 25 reactions and no more than one comment. Nearly three-quarters of cited authors were frequent posters, defined as five or more posts in four weeks. Just under half had 2,000 or more followers, which is to say just over half did not.

Read that against a normal social dashboard and the two are almost unrelated. A post with twenty reactions is a bad day in most B2B reporting. It is also, per this data, an entirely typical citation. Whatever the engines are selecting for, it is not the thing the feed algorithm selects for.

What they do appear to select for is structure and substance, and here the two studies agree closely. In Meltwater's top-cited articles, 100% used bulleted or numbered lists, 92% had clear headings, 75% named specific entities, and 67% carried quantitative data; half used an explicit comparison framework. Semrush found 54% to 64% of cited posts were sharing knowledge or practical advice rather than promoting anything, that cited articles clustered at 500 to 2,000 words and cited posts at 50 to 299, and that about 95% of cited content was original rather than reshared. Semrush also measured semantic similarity between cited LinkedIn content and the answers it fed at 0.57–0.60, above Reddit's 0.53–0.54 and well above Quora's 0.435, which suggests engines are lifting from LinkedIn more directly than they paraphrase from other community sources.

Those are the same properties that make any page extractable. The LinkedIn-specific news is that they work inside a social feed, where nobody writes that way by default.

On format the studies flatly contradict each other. Meltwater puts text posts at 72% of citations with articles at 12%; Semrush puts articles at 50–66% and feed posts at 15–28%. Different corpora, different engine mixes, no reconciliation offered by either. Anyone telling you to choose posts over articles or the reverse is picking a study, not reading the evidence.

The per-engine spread is wide enough to need an explanation: ChatGPT Search cited LinkedIn in 14.3% of responses, Google AI Mode in 13.5%, and Perplexity in 5.3%.

Nobody has published a confirmed mechanism, but the access route is a reasonable candidate. LinkedIn's robots.txt opens with a notice that automated access without express permission is prohibited and points would-be crawlers at a whitelisting address. It names dozens of specific crawlers — Googlebot, Bingbot, Applebot, DuckDuckBot and many others — with tailored rules. It names no AI-specific crawler at all: no GPTBot, no ClaudeBot, no PerplexityBot, no Google-Extended.

So the engines with a major search index behind them have a supported path to LinkedIn's public pages, and the ones relying more heavily on their own crawling do not. That the two index-backed engines cite LinkedIn at roughly triple Perplexity's rate is consistent with that story. It is consistent with several other stories too — different retrieval preferences, different source weighting, different prompt handling — so treat it as a hypothesis that fits rather than a finding. The general principle is not in doubt, though, and it is the one covered in our guide to robots.txt and AI crawlers: an engine cites what it can reach.

Three limits, and the first one is large.

Both studies are vendor research on a channel the vendors sell into. Meltwater sells media intelligence and ran this through its own GenAI Lens product; Semrush sells AI visibility tracking. Neither is disinterested about the conclusion that LinkedIn deserves more of your attention. Semrush publishes prompt counts, URL counts, engines, a date range, and a named author, which makes its figures checkable. Meltwater's write-up publishes the citation total and the engine list but no prompt count, no date range, and no category-level method, and its own press release names a narrower engine set than its blog post does. The 75/25 split everyone is quoting is the number with the least methodology behind it.

"LinkedIn is cited" is not "LinkedIn citations name your brand." Being the source behind an answer and being named in it are separate events, and on the engine that cites LinkedIn most, they come apart hardest: ChatGPT cited brands in 87% of appearances while naming them in 20.7%, per the Semrush and Kevin Indig work on ghost citations. An employee post cited without attribution to their employer is a citation that builds a person's authority and possibly nothing else.

The categories are professional and mostly B2B. Meltwater's coverage is 16 B2B categories, and LinkedIn ranked #1 or #2 in 10 of 14 it broke out — AI and data science, marketing, leadership, sales, supply chain, financial services. Nothing here says LinkedIn is cited for consumer queries, technical documentation questions, or local search. Check your own prompts before reallocating anything.

One finding does line up cleanly with the wider evidence: Meltwater found 48% of LinkedIn citations pointed at content published within the previous three months and only 12% at content over a year old. That is a much steeper recency curve than the median citation ages of five to eight months measured across the open web, which is what you would expect from a feed. LinkedIn citations are perishable in a way blog citations are not.

The uncomfortable implication is that a channel most B2B companies manage as a broadcast surface, with a brand account and a content calendar, is being read by AI engines as a collection of individual experts. The company page is in the mix. It is not the majority of the mix on the engines with the most traffic behind them.

The reframe worth carrying: on LinkedIn, AI visibility is a staffing decision before it is a content decision. You cannot schedule your way to it from a brand account, and the people who can earn it are already on the payroll and mostly not posting.

Elmo is an open-source, self-hosted AI visibility platform that runs your prompt sets across ChatGPT, Claude, Gemini, Perplexity, and Google's AI surfaces, recording every citation, mention, and competitor named alongside you. Because it stores the full cited URL rather than just the domain and the data is yours to query directly, splitting your LinkedIn citations into profiles and company pages is a query over your own results rather than a manual audit.

For the fundamentals, start with AI citations and answer engine optimization, then how to track your brand in AI search. For the wider off-site picture, see where AI citations come from; for how fast the cited set turns over, see citation volatility. For the vocabulary, see the AI search glossary.