LinkedIn Strategy for B2B: What 1,067 Posts and Comments Actually Show

Most LinkedIn engagement is invisible effort. A B2B seller I know spent six weeks commenting on every post from every account he wanted to reach. Ninety comments, a thoughtful reply within the day, and not one response from an author. He assumed he was bad at it. He wasn't. He was doing the thing 62 percent of LinkedIn does, and it doesn't work for any of them either. I know because I pulled 1,067 real posts and comments off LinkedIn and measured them, and most of what gets written on that platform is filtered out before anyone sees it.

What I measured

I extracted 486 comments and 581 posts from live LinkedIn feeds, with their reaction counts, comment counts, and format. Not a survey, not a vendor's benchmark report. The actual text people wrote and what it earned. Then I split the comments into the ones that got at least two reactions and the ones that got zero, and looked for what separated them.

One caveat before the numbers: this came off a B2B feed weighted toward marketing and technology. A feed of process engineers or procurement leads will behave somewhat differently. Where my numbers line up with the large published studies, I'll say so. Where they're only mine, I'll flag that too.

Why most LinkedIn comments are invisible

Of the 486 comments I pulled, 280 were under ten words. That's 62 percent. Of those short comments, 9 percent earned a single reaction. Not nine percent got lots of engagement. Nine percent got any at all.

Cross the ten-word line and the number jumps to between 33 and 38 percent. It stays roughly flat after that, so this isn't a case of longer always being better. There's a floor, most people are under it, and once you clear it the returns level off.

Richard van der Blom's analysis of 1.8 million posts found the same edge from the other direction: comments of fifteen words or more carry about 2.5 times the algorithmic weight, and shorter ones are rarely surfaced at all. My winners averaged 23 words. The dead ones averaged nine.

So the seller writing "Great insight, thanks for sharing" ninety times wasn't being ignored by the people he wanted to reach. He was filtered out before they ever saw it.

The name trap

This is the one that surprised me, and it's the most useful thing in the dataset.

72 percent of the comments that earned zero reactions opened with the author's first name. Among comments that did earn reactions, only 11 percent did.

You know the shape. "Sarah, great point about supply chain visibility." "Mike — this is spot on." It feels personal. It feels like the opposite of spam. And it is the single most reliable marker in my data of a comment that goes nowhere.

My read on why: opening with someone's name signals that you're addressing them rather than contributing to the conversation. It's a greeting, and a greeting adds nothing for the fifty other people reading the thread. The author sees their name, registers politeness, and moves on. Nobody else has a reason to care.

If you want to reference the author, do it mid-sentence, or don't do it at all. The comment should stand on its own without the name attached.

What high-performing LinkedIn comments do differently

The strongest positive signal in the whole set was first-person experience. Comments containing "I," "my," or "we" showed up in 34 percent of winners and 8 percent of the dead ones. A four-fold gap, wider than any other feature I measured.

The winners sound like someone who has done the work: "I've watched this conversation stall at the same point every time." The losers restate the post in different words and add nothing.

Here's the part that trips people up. Digits appeared in 49 percent of the dead comments and only 11 percent of the winners. That inverts what works in posts, where 92 percent of the high performers I measured put a number in their opening line. In a comment, a pasted statistic is what people reach for when they want to look substantive. A number only helps when it's your own result and someone could argue with it.

Which is the actual test underneath all of this: could someone reply "no, that's wrong"? If nobody can disagree with your comment, it isn't a contribution. It's noise wearing a suit.

Seven accounts, one post, one sentence

While pulling this data I ran into something I hadn't expected. On a single post about AI and management, seven different accounts had left comments in exactly the same shape:

• "Defining success becomes more important as execution accelerates."
• "Poor design becomes more costly once execution scales."
• "Attention becomes the constraint when software can stay busy indefinitely."
• "Intervention points become more important as decision processes become automated."
• "Defining done becomes more important as hands-on work disappears."

Abstract noun, becomes more important, as some process does something. Seven variations, seven different people, one post. Every one of them scored zero.

They're grammatically perfect. They sound thoughtful. And not one says anything a person could check, argue with, or act on. This is what an AI writes when you tell it to sound insightful, and it's the failure mode to watch for if you're letting a tool draft your engagement. It clears every obvious quality bar — decent length, no exclamation marks, no "great post" — and it is still completely empty.

A second pattern showed up in 13 percent of everything I collected: a first name, an elongated agreement, then a restatement. "Tosin exactlyy, building relationship equity on their feed first completely alters how they read your pitch." The stretched vowel is doing the work of sounding casual. It fools nobody.

Which LinkedIn post format performs best

On the posting side, one number dwarfs the rest.

Link posts in my sample averaged 12 reactions. Carousels averaged between 552 and 682. That's a 46-fold gap, and it holds up against the published research. AuthoredUp's analysis of 3 million posts puts document posts at 1.39x the reach and 1.30x the engagement of an average post, while Socialinsider's study of 1.3 million posts found native document posts carry the highest engagement rate of any content type on the platform. Multiple analyses put the external-link reach penalty at around 60 percent.

For most B2B teams this has a direct and slightly annoying implication. The instinct is to post a link to your new case study or capability page. That is the worst-performing thing you can do on LinkedIn. Put the substance in a carousel, and put the link in the first comment.

Carousels are multi-page PDFs uploaded as a document post. Eight to twelve slides, one idea per slide, a hook slide that promises something without giving it away, and a closing slide with one ask. The structures that work are the same ones that work in a sales meeting: a step-by-step, a problem-and-fix, or a case study that opens with the result. We made the case for that last one at length in our guide to B2B case studies, and a carousel is the cheapest possible way to publish one.

Two more things worth knowing. Text posts beat single-image posts in the current algorithm, reversing what was true two years ago. And 92 percent of the high-performing posts I measured had a number in the opening line, while only 5 percent opened with a question. The "thought-provoking question" opener that every LinkedIn guide recommends is almost absent from the winners.

When to comment on LinkedIn

This finding isn't from my data, and it might be the highest-leverage one here. An analysis of 261,137 comments found that commenting within 30 minutes of a post going up earned 391 impressions on average, against 104 for a comment left the next day. Nearly four times, for the same words.

Most of a post's distribution is decided in the first 60 to 90 minutes. A comment that lands inside that window is part of the signal that determines how far the post travels. A comment left the next day arrives after the decision has been made.

For a small B2B team, that reframes the whole activity. Fifteen minutes a day watching the twenty accounts you actually care about beats an hour a week catching up on everything. It's the same logic as the follow-up window after a trade show, which we covered in our piece on trade show ROI: the work is worth a multiple if it happens now instead of Friday.

Where LinkedIn fits in a long B2B sales cycle

LinkedIn is rarely where a considered B2B purchase gets found. Buyers still start with a search and a spec, which is why search and ungated reference material do the discovery work. LinkedIn is where you stay familiar during the twelve or eighteen months between the first search and the RFQ, alongside the email nurture carrying the same job. Treat it as the familiarity layer rather than the lead source and the effort makes sense. Treat it as a lead source and you'll quit in six weeks, like the seller at the top of this piece nearly did.

Move this week

1. Audit your last twenty comments. Count how many open with someone's name, and how many are under ten words. If you're like the corpus, it's most of them. That's your quickest fix.
2. Write comments that contain something only you know. A deal that stalled for a reason nobody expected, a number from your own pipeline, a mistake you made twice. Fifteen to thirty-five words. If it could have been written by someone who has never done the work, delete it.
3. Pick twenty accounts and check them once a day. Not a hundred. Twenty checked early beats a hundred checked late.
4. Move your next link post into a carousel. Eight to twelve slides, link in the first comment. Compare it against the last three link posts you published.
5. If you're using AI to draft, test it against one question: could someone reply "no, that's wrong"? If not, it's the seven-accounts pattern and it will earn you nothing.

None of this requires more time on LinkedIn. It requires the same time spent above the floor instead of below it. Ninety comments would have been maybe thirty with a different shape, and thirty of the right ones would have got him further than ninety of the wrong ones.

If you want a second opinion on where your team's effort is going, bring your last month of LinkedIn activity to a free 30-minute fit call and we'll tell you straight whether the problem is the writing, the format, or the timing. Book it through our contact page. Or just open your last twenty comments tonight and count the names. That number is the whole argument.