We checked 9 LinkedIn stats on our own site. None held up as written.

I sell a LinkedIn writing tool, and our own site was quoting LinkedIn stats we couldn't back up.

In September and October 2026 I checked every engagement figure on postbrander.com against its source. There were 9 specific claims. 3 were contradicted by the data, 4 had no source I could find, and 2 were real numbers stretched to say something they don't.

Here's each one, what the evidence says, and how to check the next LinkedIn stat you see.

The 9 claims

1. "Text-only posts beat single images by about 30%"

Contradicted. Every large dataset I checked says the opposite:

• Richard van der Blom's 2026 research (1.3 million posts): image plus text at 3.20% engagement, text-only at 2.00%
• Socialinsider's LinkedIn benchmarks (business pages): single images at 5.30%, text at 4.50%
• AuthoredUp: images get a 1.18x reach multiplier, text 0.88x

This one was backwards, and it's the kind of claim that changes what people post.

2. "Long-form posts get 18% more engagement"

No source. The direction is right, though. AuthoredUp's study of 372,126 posts found posts of 1,301 to 2,500 characters had a median engagement rate of 2.61-2.67%, against 2.10% for posts under 400 characters. That's about 27% more, so the made-up number undersold it.

3. "Carousels get 3.7x the engagement"

Stretched. The 3.7x traces to one vendor's dataset, from a company that sells a carousel maker. Van der Blom's 2026 numbers put document carousels at 6.60% against 2.00% for text-only: a little over 3x. Carousels really are the strongest format. The multiplier was the problem.

4. "Hashtags add 12.6% reach. Use 3 to 5."

Contradicted. I couldn't trace 12.6% to a primary source. AuthoredUp's research found hashtags had no measurable impact on reach, that 3 to 5 can slightly reduce visibility, and that more than 6 cuts it hard. LinkedIn also removed hashtag pages in October 2024.

5. "Polls get 40% more reach"

Contradicted, by time. Polls were a real reach play through 2023 and 2024. In 2026 LinkedIn demoted engagement bait, and polls now average about 0.07% engagement in van der Blom's data, the lowest of any format. Our content strategy guide was still recommending a Monday poll as a "high reach kickoff".

6. "Video gets 5x the engagement of a standard post"

No source. Native video does well: 5.60% in van der Blom's data, second only to carousels and nearly 3x text-only. But 5x isn't in any dataset I found.

7. "Video views are growing 36% year over year"

No source. I couldn't find where this came from.

8. "LinkedIn Live gets 24x the reach"

Stretched. The 24x is real, and it's LinkedIn's own figure: Live streams get 7x more reactions and 24x more comments than the same broadcasters' recorded videos. That's comments, against recorded video by the same people. It says nothing about reach compared with an ordinary post.

9. "Posts that end with a question get 2-3x more comments"

No source. It sounds plausible, which is how it survived. I found no dataset behind it.

How these got on the site

The guides went up in April 2026. I can't say exactly where each number came from, which is the problem. My guess is LinkedIn stats round-ups, because most of these appear in several of them.

Round-ups quote each other. A number gets copied, then rounded, then copied again, and after a few hops the source link has gone and only the number is left. That's how a stat ends up everywhere and traceable to no one. I'm not proud we repeated them.

What we changed

Every figure on the site now names its dataset, and the benchmarks page lists each source with the date we last checked it. Tables use one dataset at a time, so you're never comparing van der Blom's numbers with someone else's methodology in the same column.

The product changed too. Postbrander's fact-check flags figures in your drafts that you should confirm before publishing, and the writer is told never to invent statistics. Your name goes on the post, so you should be able to stand behind every number in it.

How to check a LinkedIn stat yourself

Before you repeat a stat in a post, ask 4 questions:

1. Who measured it? A carousel tool reporting that carousels win has a reason to like that result.
2. How many posts, and when? 1.3 million posts from 2026 beats a few thousand from 2022. LinkedIn's feed changed a lot in early 2026, so older numbers may describe a different feed.
3. What was measured? Reach, engagement rate and comments are different things. The Live figure above fell apart on this question alone.
4. Does a second dataset agree? The image-versus-text question has 3 independent sources pointing the same way. That's a claim you can use.

If a stat fails 2 of those, leave it out. A post with no number is better than a post with a wrong one.

Sources

• Richard van der Blom, LinkedIn algorithm research 2026, as reported by Dataslayer
• AuthoredUp, LinkedIn algorithm research and character length study
• Socialinsider, LinkedIn benchmarks
• LinkedIn Marketing Blog, What is LinkedIn Live

All checked in September and October 2026. The full set of current numbers is on the LinkedIn engagement benchmarks page.