Hey,
Writing this from my hotel room in Guangzhou.
Just wrapped a long weekend at the Bestfulfill Mastermind. A room of 7 to 9 figure DTC founders. Speaking alongside experts like Nick Shackelford, Dara Denney, Will Hughes, Chris Munch.
Operators running brands you've heard of and brands you haven't but should have.
These people are sharp.
They run paid social budgets I can't even mentally process. They know their CACs, their LTVs, their hook rates, their creative testing cadences. Performance marketing royalty.
And almost none of them use LinkedIn.
I kept asking. Some don't have a profile. Some have one from 2018 with a job title that no longer exists. Some post once a quarter and wonder why nothing happens.
Meanwhile every conversation circled back to the same thing.
Their hardest channel right now is anything that isn't paid. CPMs going up. Creators getting saturated. Attribution breaking. Everyone in the room is looking for organic plays that actually work.
LinkedIn is sitting right there.
And here's the part most of them haven't clocked yet.
LinkedIn isn't just a social platform anymore. It's an AI search surface. When your buyer types a question into ChatGPT, there's a real chance LinkedIn content is shaping the answer they get back.
So I want to walk you through what I learned digging into the data this week, because it changes the whole game for how you should be thinking about your profile and your posts.
First, the receipt
Semrush analyzed 89,000 LinkedIn URLs cited by ChatGPT, Perplexity, and Google AI Mode earlier this year.
LinkedIn came in #2 on the list of most-cited domains. Ahead of Wikipedia. Ahead of YouTube. Ahead of every major news publisher.
11% of AI responses pull from LinkedIn on average. On ChatGPT Search and Google AI Mode, it's 13 to 14%.
Scrunch puts the volume at roughly 8 MILLION LinkedIn citations PER WEEK in the US, just for commercial prompts (the exact ones your buyers run when they're researching purchases). And that volume is growing 13% month over month.
Translation: when your buyer asks ChatGPT "who's the best agency for X" or "how do I solve Y", LinkedIn content is in the room.
The only question is whose. Right now, almost certainly not yours.
The thing that broke my brain
Scrunch ran a separate study on 12,000 LinkedIn posts ChatGPT considered as sources.
Here's the finding I can't stop thinking about: reaction count has near-zero effect on whether a post gets cited.
A post with 100 reactions and a post with 10,000 reactions get cited at the same rate.
Which means the whole "I can't post on LinkedIn because I don't have a big audience" excuse is dead. The audience isn't the gating factor. The content is.
Semrush backs this up with even sharper numbers. The median post that gets cited by AI has 15 to 25 reactions. One comment. That's it.
About 75% of cited authors post 5+ times in a four-week window. And people with fewer than 500 connections beyond their immediate network get cited at the same rate (sometimes more) than people with bigger followings.
Frequency and depth beat virality. Every time.
So if you've been waiting until you have "enough followers" to start posting, stop waiting. That's not the variable.
What actually gets you cited
Scrunch tested 21 different content patterns. Three moved the needle hard.
Technical details: +77% citation rate. Specifics. Numbers. Architecture. Tools. Metrics. The kind of detail you can only write if you've actually done the thing. And here's the kicker: technical details have zero effect on reactions. So this is pure upside for AI visibility, no downside on engagement.
Generic post | Technical post |
|---|---|
"AI is transforming how we think about data pipelines." | “After running 200+ Airflow DAGs at scale, the three failure modes we see most are executor memory pressure during backfills, cross-DAG dependency deadlocks, and sensor timeout cascades on S3 event triggers.” |
Named entities: +33% citation rate (+5% reactions). Naming specific companies, people, products, frameworks in plain text. You don't have to @-tag anyone. ChatGPT reads the text of the post, not the LinkedIn mention graph. Just type the names.
Vague post | Named post |
|---|---|
“There are several great project management tools on the market." | "Notion vs. Linear vs. Asana: how their automation engines actually differ for teams over 50." |
Topic specificity: +18% citation rate (+13% reactions). Niche beats broad. "Embedded lending unit economics for vertical SaaS serving auto dealerships" beats "the future of fintech" every time. This is the only sweet spot tactic that helps both AI and humans equally.
Broad post | Specific post |
|---|---|
"The future of fintech is exciting." | "Embedded lending is changing unit economics for vertical SaaS platforms serving auto dealerships." |
And two things that hurt:
Unicode formatting: -58% citation rate. Those 𝗯𝗼𝗹𝗱 and 𝘪𝘵𝘢𝘭𝘪𝘤 characters everyone uses on LinkedIn? They're not real letters. They're math symbols that look like letters. ChatGPT literally can't read them.
This is a ChatGPT-specific issue (Perplexity and Google AI Mode handle it fine), but ChatGPT is the biggest AI traffic source you have. The penalty also applies to your name, your headline, your About section, and any other page ChatGPT crawls. So if your LinkedIn profile is dressed up in unicode, it's invisible to ChatGPT search.
Link-in-comments: -31% citation rate for the post itself. The post becomes a pointer instead of a source. ChatGPT skips it.
But here's the nuance most people miss: the URL you drop in the comments gets cited 47% of the time. That's 2x the baseline rate. So if your goal is to drive AI to your own website or article, link-in-comments can be a deliberate trade. Sacrifice the post's citation odds to massively boost the linked source. Just know what game you're playing before you do it.
Source: scrunch.com/blog/linkedin-posts-robots-cant-resist (bottom of article)
Two more things worth knowing before we get tactical.
LinkedIn articles (the long-form ones, not feed posts) make up 50 to 66% of cited LinkedIn content. The sweet spot is 500 to 2,000 words. For feed posts, it's 50 to 299 words.
Reshares barely register at 5%. Original content wins, every time.
And for founders specifically, where you post matters as much as what you post.
Perplexity cites Company Pages 59% of the time. ChatGPT and Google AI Mode flip it: individual creators get cited 59% of the time on both.
Translation: your personal profile is the asset. Not just the company page.
Do these five things this week
This is the part where most newsletters get vague.
I'm not going to do that.
Open LinkedIn in another tab and do these now.
↓
1. Strip unicode out of your profile and your last 5 posts.
ChatGPT can't read 𝗯𝗼𝗹𝗱 or 𝘪𝘵𝘢𝘭𝘪𝘤 unicode characters.
Posts that use them are 58% less likely to get cited. The penalty also hits your name, headline, and About section.
While you're in there, rewrite your headline in plain text:
[What you do] for [who you do it for].
Two minutes total. Highest-leverage two minutes you'll spend this week.
2. Add real technical detail to every post.
Posts with specifics get cited 77% more often.
This is the single biggest finding in the entire Scrunch study. Not close.
And the kicker: technical details have zero negative effect on reactions. So this is pure upside for AI visibility, no downside on engagement.
Bad: "We help brands run better Meta ads."
Good: "We tested 340 creative variants across 8 DTC brands in Q1 using Motion. The pattern: UGC ads with founder voice-overs outperformed studio-shot creative by 2.3x ROAS, but only when the hook landed in the first 1.2 seconds. Past that, the gap closed within 48 hours of launch."
Names. Numbers. Tools. Methods.
That's the citation-friendly format. It's also the format your actual buyers want to read.
3. Name specific companies, people, and tools in plain text.
Posts with named entities get cited 33% more often. They also get 5% more reactions.
You don't need to @-tag anyone. ChatGPT reads the words of the post, not the mention graph. Just type the names.
"Notion vs Linear vs Asana" beats "the top project management tools."
"Working with Semrush taught me X" beats "I worked with a major SaaS brand."
4. Write about one specific niche, not a broad trend.
Topic specificity is the only sweet-spot tactic that helps both AI (+18%) and humans (+13%).
"Embedded lending unit economics for vertical SaaS serving auto dealerships" beats "the future of fintech."
The narrower your topic, the more matchable your post is to the specific questions your buyers are asking AI tools.
5. Publish one LinkedIn article this month.
Not a post. An article.
Long-form articles make up 50 to 66% of all LinkedIn content cited by AI. Sweet spot: 500 to 2,000 words. Original content only.
Almost nobody in your space is publishing LinkedIn articles right now. Runway is wide open.
If you already have blog posts on your site, convert one into a LinkedIn article. Highest-leverage move you can make this quarter.
The brands and founders who figure this out in the next 12 months will look up in 2027 and realize they got a 3-year head start on their competitors.
The ones who keep ignoring it will pay for that distribution later. With ad spend. Or by being absent from the answers their buyers are now asking AI.
If even 9-figure operators are sleeping on this, the runway for everyone else is wide open.
Hit reply and tell me what you changed. I'm reading every email between flights this week.
D.
P.S. The founders and operators putting this into practice are inside The LinkedIn Engine.
4 weeks. Live calls. Homework. The actual content flywheel built with you, not just explained at you.
Founding rate locks in before we hit 100 members: https://www.skool.com/the-linkedin-engine