LinkedIn Is Now the Second Most Cited Source in AI Search. Your Posting Cadence Is an AI Visibility Strategy

For most of its life, the company LinkedIn page has been a box to tick. Post the press release, share the webinar, let it go quiet around month six. Nobody on the leadership team noticed, because nobody could point to a deal that came from it.

That calculation changed in 2026, and the reason is not LinkedIn’s algorithm. It is what the AI assistants your buyers now use decided to read.

The data

In March 2026, Axios reported that LinkedIn was becoming a top source for AI chatbot answers. The numbers behind that have come from several directions since.

Semrush analysed 325,000 prompts across ChatGPT Search, Google AI Mode and Perplexity and found 89,000 unique LinkedIn URLs cited in the answers. By that count LinkedIn is the second most cited platform in AI search, behind Reddit.

Profound, which tracks what AI engines cite over time, found that LinkedIn posts, long-form articles and newsletters account for about 35 percent of all LinkedIn citations within ChatGPT, with profiles at about 14.5 percent. The share for posts and articles rose from 26.9 percent in November 2025 to 34.9 percent by February 2026, eight points in three months.

Meltwater’s AI search visibility tracker for August 2026 recorded LinkedIn citations growing 24.8 percent month over month, moving LinkedIn from fifth to third place across the eight AI platforms it monitors.

These are three vendors with three methodologies, and LinkedIn itself has published its own, larger multipliers, so the exact shares will keep moving. The direction across all of them is the same. When a buyer asks an assistant which vendors to look at, how a category works, or what a company is known for, the answer is increasingly built from what has been posted on LinkedIn. Not the company’s website. The posts.

Why posts, specifically

AI search engines are looking for the same things a careful human researcher looks for: specific claims, from identifiable people and organisations, published recently, repeated across sources. We wrote about the mechanics in how AI search chooses its sources.

A LinkedIn post ticks those boxes in a way most web pages do not. It has a named author with a title and a company. It has a date. It is usually about one thing. It is often a concrete claim, a number, a worked example, an opinion with a reason attached. And it exists alongside hundreds of other posts from the same person and company, which gives the engine the repetition it needs to treat the source as credible on a topic.

A profile, by contrast, is a static list of jobs. It gets cited for “who is this person” and little else. The content is what gets cited for “what does this company think” and “who should I consider”.

That is why the shift from profiles to posts in Profound’s data matters. The engines are not learning that LinkedIn exists. They are learning to read the feed.

What this does to the posting cadence conversation

Until now, consistency on LinkedIn was argued about in terms of reach. Post three times a week and the algorithm rewards you, post in bursts and it does not, the evidence is mixed and vendors disagree. We take that argument seriously in how often should a B2B company post.

AI citation changes the argument, because it does not care about reach. A post that got forty impressions on the day it went out is still a dated, attributed, specific claim six months later, and the engine will find it when a buyer’s question matches it. The post is not a broadcast that expires. It is a source that accumulates.

Which means the question is no longer “how many impressions did we get this week”. It is “how much citable material have we published on the topics we want to be known for, and is it still growing”. A company that has posted one substantive thing a week for a year has fifty sources. A company that posted daily for six weeks in the spring and then went quiet has thirty, all from one month, on whatever was on the team’s mind that month.

The problem is that most pages go quiet

One analysis of 847 B2B company pages by Ligo found that most are abandoned within six months: about a quarter never post after setup, and nearly half stop posting somewhere between month two and month six. It is one vendor’s sample, but anyone who has looked at the company pages in their own category will recognise the shape.

Pages do not go quiet because the team stopped believing in LinkedIn. They go quiet because every post is a blank page, a request to someone for an image, and an approval thread, and in week seven something more urgent arrives. We wrote about this as a production failure rather than a discipline failure in why your company page went quiet by month six.

The AI citation data turns that failure into a measurable cost. Every quiet month is a month in which the engines found your competitor’s posts and not yours. And because citations accumulate, the gap compounds: the company that kept posting through the quiet months is now the one the assistant names.

What to post so it gets cited

Not everything on LinkedIn becomes a source. Five kinds of content do.

A concrete claim with a reason. “Most of our customers’ first campaign ships in under a week, because the images are made after the copy is approved” is citable. “Excited to share some thoughts on marketing velocity” is not.

A number you can stand behind. Benchmarks, before-and-after measurements, a count from your own data. The engines quote numbers because buyers ask for them. If you cite someone else’s number, name them, the way this post does.

An answer to a question a buyer would actually ask an assistant. “How long does it take a three-person marketing team to launch a campaign” is a question people type. Write the post that answers it.

A long-form article or newsletter, monthly. Profound’s data and LinkedIn’s own both show articles punching above their weight in citations. One a month, on a topic you want to own, is enough.

A named person saying it. Posts from a founder or a marketing leader get cited with a name attached, which is what the engines are looking for. The company page carries the proof and the legitimacy. The relationship between the two is in founder profile vs company page.

What does not get cited: engagement bait, polls, reposts with “great insights”, and anything so generic it could have come from any company. There is nothing in those to quote.

Making the cadence survivable

The cadence that gets cited is a substantive post every week and an article every month, sustained for years. That is not a lot of content. It is a lot of showing up, and showing up is what fails.

The way to make it survive is to stop treating each post as a blank page. In Marqeable, the week’s posts are drafted from a brief, in your voice, with the claims and numbers you have approved for the company. The review agent checks the language, the brand and the claims before anyone else reads it. A person approves. The image is made after the words are final, inside your brand look, so sixty posts look like one account, a problem we covered in AI images for LinkedIn posts. The approved posts sit on the marketing calendar beside the campaigns and the launches, so the week’s LinkedIn content is visible in the same plan as everything else.

The person’s job becomes a weekly approval, in minutes, not a weekly blank page. That is the difference between a page that is still posting in month eighteen and one that stopped in month six.

Measuring whether it is working

Reach is the wrong number for this. Three better ones.

1. Citable posts published, per month, on the topics you want to be known for. Count only the posts with a specific claim, number or answer in them.
2. Whether assistants name you. Ask ChatGPT, Perplexity and Google AI Mode the questions your buyers ask, monthly, and record whether you appear and what they cite. The tooling for this is in the best AI visibility tools.
3. Self-reported source on your demo and contact forms. “How did you hear about us” with “an AI assistant” and “LinkedIn” as options. It is imprecise and it is the only attribution that survives a feed that suppresses clicks.

The broader playbook for being recommended by AI search is in how to get your SaaS recommended by ChatGPT, and the difference between the three disciplines in AEO vs SEO vs GEO. LinkedIn cadence is the part of that playbook most companies already have an account for and are not using.

Frequently asked questions

Does ChatGPT cite LinkedIn posts?

Yes. Semrush’s analysis of 325,000 prompts across ChatGPT Search, Google AI Mode and Perplexity found 89,000 unique LinkedIn URLs in AI answers, making LinkedIn the second most cited platform after Reddit. Profound’s data shows LinkedIn posts, articles and newsletters make up about 35 percent of LinkedIn citations in ChatGPT, with profiles at around 14.5 percent.

What kind of LinkedIn content gets cited by AI search?

Substantive, specific content: long-form articles and newsletters, posts that state a concrete claim or number, and posts that answer a question a buyer would ask an assistant. Engagement bait and generic posts do not get cited because there is nothing in them to quote.

How often should a company post on LinkedIn to be cited by AI?

Consistently, which matters more than the exact number. AI search systems draw on recent, repeated, specific content from a source. A weekly cadence of substantive posts and a monthly article sustained for a year builds far more citable material than a burst of daily posts followed by silence.

How does Marqeable help with LinkedIn cadence?

Posts are drafted from your brief in your voice, the review agent checks claims and brand, a person approves, and the posts sit on the marketing calendar with the images already made in your brand look. Cadence becomes a weekly approval rather than a weekly blank page, which is what keeps a company page from going quiet.

The bottom line

LinkedIn is now the second most cited source in AI search, and the citations are coming from posts and articles, not profiles. A company that publishes one substantive post a week and one article a month is building the sources the assistants will quote. A company whose page went quiet in month six is building its competitor’s advantage.

Cadence fails on production, not on belief. Draft the week from a brief, review once, make the image after the copy, approve in minutes, and keep the posts on the same calendar as everything else. That is how a page is still posting in year two, which is now how a company gets named when a buyer asks an assistant who to consider. More leads generated, and more of them won as customers.

See each station of the factory: build campaigns from one brief, review and approve content, run follow-up with automations, answer visitors with website chat, and tie revenue to the exact message with attribution.

Marqeable runs your campaigns, answers every visitor, text, and email in seconds, and turns them into booked jobs and meetings - even at 9pm on a Saturday. We’re in private beta with a small early cohort. Get early access