KEY SUMMARY
Across 100 of my own LinkedIn posts, the type of post decided the result more than the topic did. Human posts, about real people and real moments, drew the most reactions, with a median of 20. Proof posts, showing results or a guest's insight, drew 14. Thesis posts, my own opinions, drew 6. But the seven posts that asked readers for something, two offers of a free playbook and five asks for help, produced 58% of all the comments. A founder's feed needs all four jobs: human posts to be known, thesis posts to be understood, proof posts to be believed, and an occasional offer to start the conversation. Judge each by its own number.
I pulled my last 100 LinkedIn posts, March to November 2026, and tagged each one by the job it was doing. Not the topic. The job. Was it an opinion, a result, a moment with a real person, or a request?
Then I lined the tags up against the reactions and comments. The reactions went one way. The conversations went the other. And the posts I'd have cut if I judged by reactions turned out to be the only ones that started conversations.

Reactions and conversations are different jobs. No post type does both.
What the hundred posts did
The tagging is mine, so take it as an operator's count, not a study. I used four jobs, plus a bucket for the 11 posts that didn't fit any of them, such as a bare link or a one-line reaction to a model launch. LinkedIn only shows impressions to the account owner, so this table is reactions and comments, the numbers anyone can see.
Job | Posts | Median reactions | Total comments | Comments per reaction |
|---|---|---|---|---|
Human | 29 | 20 | 59 | 0.08 |
Proof | 26 | 14 | 46 | 0.10 |
Thesis | 27 | 6 | 30 | 0.12 |
Offer and asks | 7 | 12 | 195 | 0.60 |
Four things in that table changed how I plan a week.
Human posts carry the reactions, by a wide margin. The median human post got more than three times the reactions of the median thesis post. Half of the top ten posts by reactions were human: the day I recorded with Kushagra Tiwari, wrapping a recording with Rahul Wadhwa, another with Nitya Chidambaram, a reflection four months after leaving my job. Only one thesis post made the top ten. The mechanism is not mysterious. A post about a person gets shared by that person, and by the people who know them.
Thesis posts are the quietest on reactions, and I'm keeping them anyway. Twenty-seven opinion posts, a median of 6 reactions. If I judged by that number alone, I'd have stopped writing opinions in April. I'll come back to why that would be a mistake.
Asking is the entire comment engine. Seven posts out of a hundred asked the reader to do something: two offered a free playbook for a comment, five asked for help, from finding a WordPress developer to voting in a pitch competition. Those seven produced 195 of the 334 comments across all hundred posts. One of them, the WordPress ask, drew 147 comments on its own. Strip that outlier out and the other six still produced a quarter of all comments from 6% of the posts.
The comment-for-a-playbook post is the cleanest example. It got 17 comments and 10 reactions. The follow-up a week later, reporting what happened when people asked for it, got 30 comments. One offer created a two-post conversation that outperformed every opinion I wrote that quarter.
Format moves the number too, but less than the job does. Video posts had a median of 15 reactions, image posts 10, plain text 8. Worth knowing. Not worth more than picking the right job for the post.
Why the standard advice gets the mix wrong
Most frameworks you'll find for founder content look alike, and they look a lot like mine. Koka Sexton runs authority, connection and conversion. Averi rotates strong opinion, specific story, tactical teardown and industry observation across Monday, Wednesday and Friday. The categories are fine. What none of them show is a number per category from real posts, and that's where the advice drifts.
The standard advice puts opinion first and most: one to two authority posts a week, the personal stuff once. In my hundred, opinion was the weakest post by every visible measure. If you follow that advice and watch your analytics, you'll conclude it isn't working and quit. The honest version is that thesis posts are a long bet that the visible numbers can't show.
The standard advice also treats the offer as a conversion step at the end of a funnel, to be used sparingly because asks "signal desperation". In my hundred, the ask was the only thing that reliably made anyone type a reply. That changes what it's for. An offer post isn't the end of a sequence. It's the one post a month that turns silent readers into people you can talk to.
And the standard advice says three to four posts a week, sometimes more. I post three. Across the months where I did, the numbers held. In August I posted four times all month. September's eight thesis posts were the quietest run in the whole set. I can't prove the gap caused it, but I've stopped letting a month go thin.
Thesis: the quiet post that does the most work
A thesis post is your opinion on how your market works. Not a news summary, not a tips list. A take you'd defend in a room.
Six reactions a post sounds like failure. It isn't, for two reasons.
The first is who reads them. In the 2025 Edelman and LinkedIn B2B Thought Leadership Impact Report, 86% of the hidden buyers in a deal, the finance, operations and procurement people who shape a purchase from behind the scenes, said they favour perspectives that challenge their assumptions. Those people don't react. They read, and they remember who had a view.
The second is what a thesis post does to every other post. When a stranger checks my profile before accepting a connection request, the human posts tell them I'm real. The thesis posts tell them what I think. Nobody takes a meeting with someone who has no opinions, however many photos they post.
The best thesis post I've seen this year came from Nitin Bhadauria, co-founder of Lucidity, on the day a big vendor launched a feature in his space. It makes four moves:
It hooks on the news. "Cisco just shipped the mistake we spent ten years making." People already know the story, so they stop.
It gives a read nobody else is giving. Everyone else posted the launch. He posted why it won't work.
It says why he's allowed to say it. One line of credibility. No résumé.
It ends on a real question. Not "thoughts?", but one his buyers have actually argued about.

A thesis post that ends on a real question is the only kind that gets a reply.
That last move matters more than I thought. The thesis posts of mine that drew any comments were mostly the ones written as a doubt rather than a verdict, such as admitting I'd developed an aversion to AI-generated content and didn't know why, or saying I no longer knew which of my posts were working. The confident verdicts got read and left alone.
Proof: the post that makes the opinion believable
A proof post shows what happened when the work got done: a result, a test, a before and after, or a guest on the show saying something you can't get elsewhere.
My proof posts sat in the middle on reactions, 14 at the median, with podcast posts doing the heavy lifting. The ones that did best led with the guest's name and the story behind the episode, such as watching Kushagra's run to the Clay World Championship before he came on the show. Clips with one sharp claim and no person in the first line sat lower. The guest carries the post; the claim carries the comment.
Two rules I hold to. Numbers carry proof, so use them, but only numbers you can stand behind. If a stat is a vendor's estimate, say so or leave it out. And credit generously. If the idea was a guest's, say so. A proof post that quietly takes a guest's idea as your own is a thesis post with a lie in it.
Human: the post that makes you worth talking to
A human post is a real moment with a real, named person: an event you spoke at, a book someone sent you, a message that made your week, a recording day.
It's tempting to dismiss these as soft. They were the top of my table by a distance, and they're the reason a stranger accepts a connection request. The rule that holds is to name the person. My best human posts were about Kushagra, Rahul, Nitya, Shubham, Mohita. The ones about me alone did well only when they were honest about something unfinished, like four months in with nothing to show yet.
What breaks a human post is turning it into a lesson. No tidy moral at the end, no scene-setting at 3 AM, no "here's what this taught me about leadership". Say what happened, name the people, stop.
Offer: the post that asks
An offer post gives something useful away, such as a playbook, a template or a benchmark, and asks for a comment in return. In my hundred it's the only post type that made people type.
It will have low reach. It will have middling reactions. None of that is the job. The job is to find out who, among the people reading quietly, wants to talk. Seventeen people asked for the outbound playbook in the comments of one post. Those are seventeen conversations I'd never have had from a thesis post with ten times the reach.
Three things make it work:
Ask for a comment, not a DM. Comments are public, and every one pushes the post to more people. A DM is invisible. Take it to DM after they comment.
Reply to everyone the same day, and send the thing by DM, not a public link. The DM is where the conversation starts. If they're not connected yet, reply "Let's connect so I can share in DM", and your network grows from people who asked.
Keep it to every two or three weeks. The ask works because the other three post types built the goodwill. An offer every week spends it.
The five asks for help in my hundred did even better than the offers on comments, which is worth a pause. People like being asked. A founder who asks for a recommendation, a vote or an introduction once a month isn't weaker for it. They've given 200 people a reason to show up in the comments.
The week I run now
Three posts a week. Monday a thesis, Wednesday a human post, Friday a proof post, and every two or three weeks the offer takes Friday's slot.

One hour on Saturday for three ideas. Ten minutes a day for comments.
The hundred posts changed three details inside that rhythm. The human post gets the video if I have one, because video posts drew the most reactions and human posts are where reach does the most good. The thesis post ends on a question I actually hold, not a verdict. And the offer is now a fixed slot, not something I remember to do when a playbook happens to be ready.
The idea for each post is still mine, worked out in an hour on Saturday from the week's conversations. AI drafts the second version. It can't think of the first, and buyers can tell when it tried.
I'd have cut the opinions in April if I'd trusted the reaction count. They're the reason anyone reads the rest.
The skeleton for each post type, the full offer-post structure and how to build a writing skill that drafts in your voice are in The B2B LinkedIn Playbook.
FAQs
What should a B2B founder post on LinkedIn?
Four kinds of post, each with its own job: opinions on how your market works, results and guest insights, real moments with named people, and an occasional offer that asks for a comment.
Which type of LinkedIn post gets the most engagement?
It depends on the kind of engagement. In a tagged set of 100 posts, posts about real people drew the most reactions, while posts that asked readers for something drew most of the comments.
How often should a founder post on LinkedIn?
About three times a week, kept up for months. Consistency across months matters more than volume in any one week.
Do opinion posts work on LinkedIn if they get few reactions?
Yes. Buyers who shape purchases from behind the scenes read opinions without reacting, and your recent opinions are what a prospect sees when they check your profile before accepting a connection.
Should a lead magnet post ask for a comment or a DM?
A comment. Comments are public and push the post to more people; a DM is invisible. Move the conversation to DMs after someone has commented.
Can AI write a founder's LinkedIn posts?
It can draft the second version. The idea has to be yours, because readers recognise generic AI posts quickly.
AUTHOR
Builds and runs go-to-market motions for B2B companies rather than advising from the sidelines. First hire and zero-to-scale operator across HR, legal, and sales tech, with a track record of taking businesses from early customers to 200-plus in North America.

How Many People to Invite to a B2B Dinner, and What It Really Costs

AI Didn't Break B2B Sales. It Exposed It

