In August 2026, LinkedIn published a playbook explaining how businesses and creators can make their content more likely to be picked up and cited by AI tools. They call the framework the Credibility Stack, and the short version of the advice is that you should publish educational content, structure it clearly, include real numbers and named examples, and keep at it consistently.

If that sounds like fairly ordinary content advice, that’s because it largely is. What makes it worth paying attention to is the research sitting behind it, and one finding in particular that changes the maths for smaller businesses and independent professionals.

I want to walk through what the data actually says, point out a couple of places where the studies contradict each other, and be honest about why LinkedIn is publishing this in the first place.

Why This Is Happening At All

More people are asking AI tools questions instead of typing them into a search box. When someone asks ChatGPT which project management tool suits a ten person team, or asks Gemini how to evaluate an SEO consultant, the tool assembles an answer from sources it retrieves and often cites where the information came from.

That creates a new question for anyone marketing a business. Not just whether you rank in search results, but whether you turn up inside the answer itself.

LinkedIn has become a significant source for those answers, at least for professional and business queries. Two separate studies published this year both put LinkedIn near the top of the list of cited domains.

Semrush analysed 89,000 LinkedIn URLs that appeared in AI-generated responses, drawn from 325,000 prompts across ChatGPT Search, Google AI Mode and Perplexity. In their dataset, LinkedIn was the second most cited domain, appearing in around 11% of AI responses on average.

Meltwater, working with LinkedIn, analysed 9.5 million citations across six AI platforms including ChatGPT, Google AI Mode, Google AI Overviews, Gemini, Copilot and Claude. They also placed LinkedIn second, behind YouTube.

The reason is not mysterious. LinkedIn content comes attached to a professional identity. A post has a name on it, a job title, a company, an industry, and a visible history of activity. When someone with relevant credentials explains something specific about their own field, that gives a retrieval system reasonable grounds to treat it as a useful source.

What Actually Gets Cited

This is the useful part, because the pattern is fairly consistent across both studies.

Educational content, not promotional content

Semrush found that somewhere between 54% and 64% of cited LinkedIn posts were focused on sharing knowledge or practical advice. On Google AI Mode it was close to two thirds. Content promoting a product or service was cited too, but noticeably less often.

The pattern makes sense when you think about what the tool is doing. Someone asked a question. The system is looking for something that answers it. A post explaining how a process works answers a question. A post announcing that your company has won an award does not.

Structure that can be parsed

Meltwater looked at the top cited LinkedIn articles in their sample and found a recognisable shape:

Feature of top-cited LinkedIn articles	Share
Bullet points or numbered lists	100%
Clear section headings	92%
Named companies, tools or entities	75%
Hard numbers and data	67%
Comparison or evaluation frameworks	50%
“How to choose” guidance	33%

That’s not a coincidence. Those are the features that let a system pull out a specific chunk and use it to answer a specific question. A long unbroken essay might be excellent, but there’s nothing in it that can be lifted cleanly.

Formats that match how people ask

The formats appearing most often in Meltwater’s top cited content were:

Content format	Share of top-cited content
Best-of listicles	54%
Side-by-side comparisons	50%
“How to choose” guides	33%
Educational explainers	17%
Thought leadership backed with data	8%

Pure opinion, on its own, was much less likely to be cited.

This maps almost exactly onto how people phrase questions to AI tools. They ask what’s best, how two things compare, how to pick between options, what criteria matter, and what could go wrong. Content that directly supports a decision is easier to reuse in an answer than content that circles a topic thoughtfully.

Original and recent

Meltwater found 72% of cited content was original rather than reshared, and 48% had been published within the previous three months. Content older than twelve months made up only 12% of citations.

Semrush found something even starker on originality. Roughly 95% of cited posts in their dataset were original, with reshares accounting for about 5%.

Sharing other people’s posts has its uses, but it does very little for this particular goal.

The Finding That Matters Most If You’re Small

Here’s the part I’d underline for most of the businesses and professionals I work with.

Follower count barely moves the needle.

Meltwater found that 51% of cited creators had fewer than 10,000 followers. Accounts in the 1,000 to 10,000 range accounted for the largest single share of citations at 40%.

Semrush went further. In their analysis, individuals with fewer than 500 followers were about as likely to be cited as individuals with more, sometimes more so. The median cited post had somewhere between 15 and 25 reactions and no more than one comment.

Read that again if you’ve ever felt that LinkedIn is a game you can’t win because you don’t have an audience. The posts getting pulled into AI answers are, for the most part, not the viral ones. They’re quiet, specific, useful posts from people who know their subject.

What did correlate with citation was consistency. Semrush found around 75% of cited post authors were frequent posters, meaning more than five posts in the previous four weeks. The likely explanation is unglamorous: the more relevant content you have published, the more opportunities exist for something of yours to be the best available answer to a given question.

For a solo consultant, a lawyer building a practice, or a small business owner, this is genuinely good news. Reach has always favoured whoever already had reach. Relevance is a fairer contest.

Where the Studies Disagree

I’d rather flag this than pretend the picture is tidier than it is.

On format, the two studies contradict each other. Meltwater found that plain text posts made up 72% of cited content, with articles at 12% and video at 11%. Semrush found close to the opposite, with articles accounting for 50% to 66% of cited LinkedIn content and feed posts making up 15% to 28%. LinkedIn’s own playbook sides with articles, saying long form performs better than feed posts.

They may be measuring slightly different things, over different periods, across different platform mixes. But the honest reading is that nobody has a settled answer on whether posts or articles work harder, and you probably shouldn’t bet everything on one format.

On who gets cited, it depends which tool you mean. Meltwater reported that 75% of LinkedIn citations came from individual members rather than Company Pages. Semrush found that split reverses depending on the platform: Perplexity cited Company Pages 59% of the time, while ChatGPT Search and Google AI Mode leaned towards individual creators at 59% each.

So the popular summary of this research, that people beat brands, is too simple. It’s true on some platforms and false on others. If you run a business, the sensible response is to do both rather than pick a side.

And both studies were produced in collaboration with LinkedIn. That doesn’t make the numbers wrong, and both firms published their methodology. But it’s worth holding in mind that the organisations telling you LinkedIn is essential all have a commercial interest in you believing it.

The Bigger Caution

LinkedIn is not publishing this playbook as a public service. It benefits enormously if businesses come to see LinkedIn as the main route into AI discovery, because that means more posting, more time in the app, and more advertising spend.

The risk of taking the advice too literally is dependency. AI tools could shift how much weight they give LinkedIn. LinkedIn could change how its content is surfaced or licensed. If your entire discoverability strategy sits on a platform you don’t control, you’re exposed to decisions made by two sets of companies, neither of which will consult you.

There’s a more durable way to read all of this, and it’s the part I’d actually act on.

The characteristics that make content citable are not LinkedIn features. Clear structure, a direct answer near the top, named specifics, real numbers, comparison frameworks, firsthand experience, regular updating. Those work on your own website too. Meltwater’s data showed company websites still accounted for 18.7% of all citations, which is a meaningful share.

So the sensible arrangement is that your own site holds the durable version of your expertise, the pages that fully answer the questions your buyers ask. LinkedIn extends the surface area, putting a version of that thinking where AI systems and human buyers are both already looking. One reinforces the other, and you’re not renting your entire presence.

That’s also a more comfortable position commercially, because a website you own keeps working whether or not the platform’s priorities change.

How to Actually Apply This

Here’s what I’d do, roughly in order.

Start with the questions, not the content calendar

Write down the questions your customers or clients actually ask before they hire someone like you. Not topics, questions, phrased the way a person would type them into an AI tool.

For a chartered accountant that might be how to choose between the old and new tax regime, or what compliance a small company misses in its first year. For an interior designer it might be how much a two bedroom renovation costs in a particular city, or how to evaluate a contractor’s quote.

Aim for twenty five to fifty of these. That list is your content plan.

Answer one question properly per piece

One question, one piece. Put the direct answer near the top rather than building to it. Then add the context, the criteria, the examples and the trade offs underneath.

Semrush found that AI responses citing LinkedIn showed high semantic similarity to the original content, meaning the answers often closely mirrored the meaning of what was written. That has a practical implication. If your explanation of your own category is vague, the vague version is what gets repeated. If it’s precise, the precise version travels.

Be specific enough to be useful

Name the tools. Name the categories. Give the number. Say what it cost, how long it took, what went wrong the first time. Generic advice is indistinguishable from every other piece of generic advice and there’s no reason for anything to pick yours.

This is also where firsthand experience matters. You’ve seen things happen that aren’t written down anywhere. That’s the most defensible content you have, and it’s the hardest for anyone else to reproduce.

Structure it so pieces can be lifted out

Headings that describe what’s in the section. Bullet points where you’re listing things. A short summary near the top. An FAQ at the bottom covering the adjacent questions people ask.

None of this requires writing like a robot. It’s the same structure a good reference article has always had, and human readers benefit from it too.

Publish on a schedule you can actually keep

The research points towards a couple of posts a week per person. I’d rather you commit to one a week and sustain it for a year than commit to five and stop in March. Consistency is doing the work here, and consistency is a function of what’s realistic for you.

Put the important version on your own site

Write the full piece where you control it, then adapt it for LinkedIn. The website version can be longer, updated over time, linked to your services, and it belongs to you.

What This Looks Like In Practice

For a solo professional, this is mostly personal. Your name and your face are the credibility signal. Publish under your own profile, write about what you actually handle day to day, and don’t worry about follower count. One clear, specific post a week about a real question your clients ask is a reasonable target.

For a small business, the useful move is to identify the two or three people who genuinely know things and support them in publishing. Not by handing them company announcements to reshare, which the data says achieves very little, but by helping them turn what they know into something readable. Your Company Page still matters, particularly given the Perplexity finding, so keep it current and treat it as a hub rather than a noticeboard.

For a larger organisation, the shift is from distribution to editorial. Traditional advocacy programmes create a message centrally and ask staff to amplify it. The approach that fits this research runs the other way: start from the questions, find the people with real experience of them, and help them publish in their own voice. The risk to watch is that everyone starts writing identically formatted content that says nothing in particular. Structure is meant to make expertise easier to find, not to sand off the person who has it.

A Note on Realism

Being cited in an AI answer is not the same as getting a client. It isn’t even necessarily traffic. Someone can read an answer that mentions you and take no action at all.

What it does give you is presence at the moment someone is forming an opinion about their options, which is earlier than most marketing reaches people. That’s worth something, but it’s worth being clear eyed about what it is.

The other thing worth saying is that most of this advice would have been good advice five years ago. Answer real questions clearly, be specific, use structure, publish consistently, share what you’ve actually learned. AI retrieval has changed who benefits from doing that, and it has made the small-but-specific approach more viable than it used to be. It hasn’t changed what good content looks like.

Frequently Asked Questions

Do I need a lot of LinkedIn followers to be cited by AI tools?

The research says no. Meltwater found 51% of cited creators had fewer than 10,000 followers, and Semrush found individuals with under 500 followers were cited about as often as those with more. Median engagement on cited posts was around 15 to 25 reactions. Relevance and consistency appear to matter considerably more than audience size.

Should I write LinkedIn articles or feed posts?

The two major studies disagree on this, so I’d do both. Semrush found articles made up the larger share of citations, while Meltwater found text posts did. LinkedIn’s own guidance favours long form articles. A reasonable approach is a longer piece that answers a question thoroughly, plus shorter posts that make one point each and link to it.

How long should the content be?

Semrush found articles between 500 and 2,000 words were cited most often, and feed posts between 50 and 299 words. Long enough to answer the question properly, short enough to stay on topic.

How often should I publish?

The research suggests roughly two to three posts per week per person, and found that around 75% of cited post authors had published more than five times in the previous four weeks. Pick a cadence you can maintain for a year rather than the one that looks most impressive on paper.

Does resharing other people’s posts help?

Very little, for this purpose. Semrush found about 95% of cited posts were original, with reshares at around 5%. Meltwater found 72% original. Resharing has social value, but it isn’t going to get you cited.

Is this only relevant for B2B?

Mostly, though not exclusively. LinkedIn ranked in the top five cited domains across fourteen business categories in Meltwater’s analysis, including technology, financial services, marketing, HR, legal, real estate and healthcare. For consumer businesses with no professional angle, your effort is likely better spent elsewhere.

Should I do this instead of SEO on my own website?

No, alongside it, and I’d weight the website more heavily. Company websites still accounted for a significant share of AI citations, and your site is the one property you actually control. The same principles apply in both places.

Does using AI to write the posts help or hurt?

The research doesn’t measure this directly. What it does show is that citations go to specific, firsthand, original content, which is precisely what generic AI output tends not to be. Using AI to structure or tidy your thinking is fine. Publishing content that contains nothing only you could have written defeats the purpose.

If You’re Trying to Work Out Where to Start

The advice in the LinkedIn playbook is reasonable. The awkward part is the gap between understanding it and running it, week after week, alongside actually doing your job.

Working out which twenty five questions matter for your business takes some thought. So does deciding what belongs on your website versus LinkedIn, getting the structure right without making everything read like a template, and keeping it going long enough for it to compound.

That’s the work I do with clients, connecting content, SEO and social so the same expertise shows up in more than one place and none of it depends entirely on a platform someone else controls.

If you’re looking at your own presence and wondering whether any of this would work for your industry, or you’d just like a second opinion on where to put your effort, that’s an easy conversation to have.

Click here and tell me what you’re working on

Happy to look at what you’re publishing now and tell you honestly whether it’s worth building on.

This article is for informational purposes only.

Deepak Vankar is a digital marketing consultant working with businesses and professionals on SEO, content, performance marketing and brand strategy.