AI Visibility on LinkedIn: 5 Studies

Picture this: the CEO of a software company is looking for a new sales partner. She asks ChatGPT for suitable providers, gets an answer with five recommendations, and clicks through the sources. Before she has spoken to anyone, her shortlist is already set.

According to a YouGov survey commissioned by the SEO agency Seokratie among 509 business decision-makers in Germany (published September 2026), more than half use AI tools several times a week to research products and service providers. Almost one in four does so daily.

This turns the question "Where does AI actually get its answers?" into a business case. If you want to show up in those answers, you need to prepare your content so that AI systems use it as a source. The established term for this is Generative Engine Optimization, or GEO. One source is playing an increasingly important role here: LinkedIn. So it's no surprise that one analysis or study after another is being published on the topic. But their numbers differ considerably. Read just one study, and you can easily draw the wrong conclusions.

That's why I compared five major data studies conducted between summer 2025 and summer 2026. Together, they analyze more than 110 million AI citations. Put their methodology and results side by side, and clear common ground emerges. From it, you can work out which approach makes sense right now.

The comparison below shows where the studies contradict each other and how to put the findings into practice.

What you'll take away from this article

• what each of the five studies measured and found, summarized in a few sentences
• why the studies partly contradict each other
• where the studies agree: the findings supported by every study that examined them
• four concrete steps to strengthen your AI visibility through LinkedIn

LinkedIn AI Citations: Five Studies, Five Perspectives

Before we get to the results, a quick look at the methodology is worthwhile. The five studies don't measure the same thing, and that explains most of the contradictions.

All five come from companies that offer software for search, AI visibility, or media monitoring. Two of them, the 2026 Semrush study and the Meltwater study, were produced in collaboration with LinkedIn. That doesn't invalidate the numbers, but it's part of the context.

I'll go through the studies in the order they were conducted. That way, you can see how the picture becomes clearer with each one.

Semrush 2025: LinkedIn Moves Into the Top Sources

How it was measured: From July to October 2025, Semrush analyzed around 230,000 prompts, which produced more than 100 million citations. The question was which domains AI systems cite most often.

Key findings: In Google AI Mode, LinkedIn was one of the most-cited domains and consistently appeared in almost 15% of answers. On Perplexity, LinkedIn ranked second behind Reddit. On ChatGPT, LinkedIn made it into the top 5 from mid-September 2025 onward.

This study doesn't reveal which LinkedIn content was cited. But it does show that by fall 2025, LinkedIn was already among the most important sources on all three platforms (ChatGPT, Google AI Mode, and Perplexity).

Profound 2025/26: LinkedIn Leads for Professional Questions

How it was measured: Profound analyzed around 1.4 million citations from mid-November 2025 to mid-February 2026, drawn from two sources. For the ranking on professional questions, a fixed set of typical professional prompts was run across six AI models: ChatGPT, Gemini, Google AI Overviews, Google AI Mode, Microsoft Copilot, and Perplexity. The trend on ChatGPT is based on real prompts that people entered there.

Key findings: For professional questions, LinkedIn was the most-cited domain across all six models. On ChatGPT, LinkedIn climbed from 11th to 5th place among the most-cited domains within three months.

Even more revealing is which LinkedIn pages ChatGPT cites. In mid-November 2025, around 34% of LinkedIn citations still went to profile pages. By mid-February 2026, that figure had dropped to around 15%. Over the same period, the share of posts and articles rose from around 27% to around 35%.

This suggests that AI systems increasingly focus on what people write and post, and less and less on what else is on their profile.

Semrush 2026: Which LinkedIn Content AI Cites

How it was measured: For this study, Semrush worked with LinkedIn. It analyzed 325,000 prompts and 89,000 cited LinkedIn URLs from ChatGPT Search, Google AI Mode, and Perplexity between January and February 2026.

Key findings: LinkedIn appears in around 11% of all AI answers, making it the second most-cited domain, ahead of Wikipedia and YouTube. On ChatGPT and Google AI Mode, the figure is around 14% each; on Perplexity, only around 5%.

Articles account for 50 to 66% of cited LinkedIn content, feed posts for 15 to 28%. The most frequently cited pieces are articles with 500 to 2,000 words and posts with 50 to 299 words. Around 95% of cited posts are original content, not reshares of other people's posts. More than half share knowledge or give advice. On ChatGPT and AI Mode, 59% of citations go to individuals; on Perplexity, by contrast, 59% go to company pages. Around 75% of cited authors post regularly: they published more than five posts within four weeks, or more than one per week on average.

Especially interesting for anyone who regularly tracks their post engagement: The typical cited post had 15 to 25 reactions and at most one comment. Just under half of the cited post authors have more than 2,000 followers. Yet authors with fewer than 500 followers were cited at least as often as those with more. If you deliver substance, you have a good chance even without a large reach.

OtterlyAI: The Largest LinkedIn-Only Dataset

How it was measured: From January to May 2026, OtterlyAI analyzed around 1.31 million citations across 161,000 LinkedIn URLs on six platforms. Of the five studies, it is the largest dataset that focuses solely on LinkedIn.

Key findings: Around 92% of citations go to content from individuals. LinkedIn articles receive around 72% of citations, posts most of the rest. LinkedIn's share of all social media citations rose from around 8% in January to around 12% in May 2026.

How many likes, comments, or hashtags a piece of content has, and how long it is, shows practically no statistical relationship with how often it gets cited. What stands out instead is how unevenly citations are distributed: The most-cited 1% of LinkedIn pages account for around 30% of all citations. Half of all cited pages, on the other hand, were cited no more than twice. In other words, a few pieces of content get cited very often, while most are cited only occasionally.

The study works with a raw and a cleaned dataset, so it reports slightly different values in some places. The figures used here are the ones OtterlyAI itself presents as its core findings.

Meltwater: The B2B Perspective

How it was measured: Meltwater analyzed 9.5 million AI citations across 16 B2B categories and published the results in May 2026. It measured each source's share of all citations.

Key findings: LinkedIn accounts for around 0.5% of all citations, ranking second behind YouTube at around 1.5%. The figures look small because AI citations are spread across a huge number of different sources, as Meltwater itself explains. In tech and SaaS, consulting, financial services, marketing, and HR, LinkedIn is among the five most-cited sources.

75% of LinkedIn citations go to individuals. In terms of format, Meltwater finds text posts clearly ahead at 72%. Articles account for 12%, videos for 11%, documents for 5%, and images for 1%. The report doesn't explain how the formats are defined, for example whether a post with text and an image counts as a text post or an image. The analysis of the 24 most-cited articles is particularly revealing. All of them use lists, and 92% use subheadings. 75% name specific companies or tools, and 67% contain hard numbers.

Meltwater's recommendations include defining three to five core topics the brand should be associated with and having in-house experts with real hands-on experience post. This means executives just as much as specialists from day-to-day operations.

For comparison: In August 2026, Meltwater analyzed around 7.3 million AI citations for consumer topics such as banking, cars, and fashion. There, only around 7% came from social media, but around 29% came from earned media and news, meaning editorial coverage such as press articles that no company paid for. In the B2B study, by contrast, platforms with user-generated content such as LinkedIn, Reddit, and YouTube together account for around 48% of citations. The categories of the two analyses don't match exactly, but the direction is clear: For B2B questions, AI relies on such platforms much more heavily than for everyday questions.

Where the Studies Contradict Each Other, and Why

Read the five studies one after another, and you'll run into a few contradictions. Most of them only resolve once you take a close look at the methodology.

How Often Does AI Cite LinkedIn: 11 Percent or 0.5 Percent?

According to Semrush, LinkedIn appears in around 11% of all AI answers. According to Meltwater, LinkedIn's share of citations is only around 0.5%. That looks like a contradiction, but both figures can be right because they measure different things. Semrush counts how many answers include LinkedIn as a source at least once. Meltwater counts LinkedIn's share of all source citations overall.

A simplified example: An AI answer lists 20 sources, one of which is a LinkedIn post. For Semrush, this answer counts as a hit, because LinkedIn appears in it. For Meltwater, LinkedIn is just one of 20 citations here, or 5%. Because AI systems cite a huge number of different websites overall, even the most-cited sources end up with small percentages in Meltwater's data. So both studies show the same thing: LinkedIn is one of the most important sources, just measured on different scales.

Articles or Posts: Which Format Wins for AI Citations?

Semrush and OtterlyAI see articles clearly ahead. Meltwater puts text posts ahead at 72%, with articles at only 12%. Part of the contradiction probably lies in what each study counts. Meltwater itself writes that articles earn 6.5 times more AI citations than regular posts. Given the format shares, this most likely refers to citations per piece.

OtterlyAI also finds that an article earns more citations per page than a post. Per piece, then, articles come out ahead. That posts still account for the larger overall share in Meltwater's data could simply be because there are far more posts than articles.

This doesn't fully resolve the contradiction, though, since Semrush and OtterlyAI also find articles ahead in total. In addition, Meltwater specifically examined B2B categories, while the other studies work with their own prompt sets and platforms. What we can say: A single article gets cited more often than a single post. Whether posts or articles generate more citations overall remains open.

Person or Company Page: Who Does AI Cite More Often?

Here, the direction is clear; only the strength varies. OtterlyAI finds around 92% for individuals, Meltwater 75%, and Semrush 59% for ChatGPT and Google AI Mode. The only exception is Perplexity, where, according to Semrush, 59% of citations go to company pages.

How Many Followers Do You Need?

According to Semrush, authors with fewer than 500 followers are cited at least as often as those with more. In Meltwater's data, 51% of cited authors have fewer than 10,000 followers. 38% have between 10,000 and 100,000 followers, and another 10% have more than 100,000. The study doesn't say how large these groups are among all LinkedIn users.

What we can say for sure: You don't necessarily need a large reach to get cited. Whether reach improves your chances, the data leaves open.

Where the Studies Agree: What Drives AI Visibility on LinkedIn

Three findings are supported by every study that examined them. If you follow them, you're most likely on the right track based on the current data.

LinkedIn Is One of AI's Main Sources for Professional Questions

This is the only point all five studies confirm. Semrush places LinkedIn in the top 5 in 2025 and in second place in 2026. Profound ranks LinkedIn first for professional questions, Meltwater second overall and in the top 5 across five B2B industries. OtterlyAI measures a growing share of social media citations.

AI Cites People With Expertise

Meltwater, OtterlyAI, and Semrush see individuals clearly ahead, with the exception of Perplexity. Semrush also shows that almost all cited posts are original content and that more than half share knowledge or give advice. Profound fits this picture: AI increasingly cites what people write and less and less their profiles.

Substance Counts, Applause Hardly Does

Likes and comments play almost no role in citations. Semrush shows this with the 15 to 25 reactions of the typical cited post, and OtterlyAI with its analysis. What does stand out is substance: The most-cited articles in Meltwater's analysis are consistently well structured and contain specific names and numbers.

Plus a Strong Signal: Consistency

According to Semrush, around 75% of cited authors published more than five posts within four weeks, or more than one per week on average. This figure comes only from the Semrush study; the others don't provide data on it. That's why I treat it as a signal rather than as a shared finding.

How to Put It Into Practice: GEO on LinkedIn in Four Steps

Four concrete steps follow from these shared findings. None of them guarantees that large language models like ChatGPT will mention your company. But they align your LinkedIn presence with what gets cited in the studies.

1. Define your topics. A simple thought experiment: For which questions should your company ideally be mentioned in AI search?
2. Make your experts visible. Have the people with real hands-on knowledge post, not just senior leadership. A company page still makes sense, especially for Perplexity, but posts from your team are more valuable.
3. Write specifically and with clear structure. Your own experiences, names, tools, and numbers, plus lists and subheadings in articles.
4. Post consistently, in both formats. At least one to two posts a week, plus an article now and then. According to Semrush, posts with 50 to 299 words and articles with 500 to 2,000 words are cited most often.

What the Data Doesn't Answer

All five studies show correlations, not causes. Whether regular posting leads to citations, or whether active authors simply have more citable content, remains open.

They are also snapshots. Profound shows how quickly rankings shift: On ChatGPT, LinkedIn climbed from 11th to 5th place among the most-cited domains between mid-November 2025 and mid-February 2026.

LinkedIn itself hasn't published its own study so far, but it contributed data to the Semrush study. Davang Shah, VP of Marketing at LinkedIn, references the Meltwater data on the LinkedIn Marketing Blog.

Sources

• Semrush: The Most-Cited Domains in AI: A 3-Month Study (Nov. 10, 2025)
• Profound: LinkedIn is the most-cited domain for professional queries in AI search (Mar. 9, 2026)
• Semrush: We Analyzed 89K LinkedIn URLs Cited in AI Search (Mar. 10, 2026)
• OtterlyAI: LinkedIn AI Search Citations Study (June 3, 2026)
• Meltwater: How LinkedIn Content Wins in AI Search (May 12, 2026)
• Meltwater: AI Search Visibility Report: August 2026
• Seokratie: YouGov survey of 509 business decision-makers (Sept. 17, 2026)
• Davang Shah, LinkedIn Marketing Blog: Why Your Brand Needs a Point of View AI Can't Replicate (Sept. 29, 2026)

What Follows From This?

AI systems primarily cite people who have something specific to say. So whoever posts on LinkedIn for your company helps shape who speaks for your company in AI answers. That affects marketing and sales alike.

At SalesPlaybook, this is exactly what we help B2B companies with. As part of our LinkedIn service, we develop the right strategies and topics and build thought leaders who can reach buyers in their ideal customer profile (ICP) before the first sales conversation even takes place. The aim is visibility both on LinkedIn itself and in ChatGPT and Co., which use LinkedIn as a source for their answers.