Here’s where I want to slow down, because this is the part that trips people up.
You’ll see plenty of stats claiming video viewership or video performance jumped significantly year over year on LinkedIn. Those numbers are usually real. People are watching more videos on the platform than they used to.
But a separate, large-scale analysis of tens of thousands of video posts found something different: video underperforms across every follower tier, and the gap widens the bigger the account. At the highest follower counts, video was the single weakest format measured.
Those two findings aren’t contradicting each other on the facts. They’re measuring different things. “Video consumption is growing” refers to total supply and watch time across the platform. “Video reach per post is falling” is a statement about what happens to your individual post when you publish it. A platform can have more total video minutes watched while the average video creator gets a smaller and smaller slice of that attention, especially once the format gets flooded with low-effort, repurposed content that wasn’t built for LinkedIn in the first place.
That’s exactly what several analysts pointed out: much of the video now appearing in feeds is recycled from other platforms, and LinkedIn’s algorithm has gotten noticeably better at distinguishing between content built for this specific audience and content that wasn’t.
Why document carousels are winning the dwell-time game
If dwell time is the metric that matters, it’s worth asking which format actually produces it.
Document posts, the PDF carousel format, consistently post the highest engagement rates of any content type on LinkedIn right now, in some analyses by a wide margin over every other format. The mechanic is straightforward once you think about it: a carousel requires the reader to physically swipe through multiple slides to see your full point. Each swipe is an additional deliberate dwell time. A video can autoplay for three seconds while someone’s thumb keeps moving, and that interaction barely registers. A carousel can’t be skimmed the same way.
That’s not a knock on video as a medium. It’s a structural difference in how each format produces (or fails to produce) the exact signal the ranking model is built to detect.
The part nobody mentions: LinkedIn has a financial reason to push video anyway
This is the angle most algorithm breakdowns skip entirely, and it’s the one that actually explains the contradiction.
In 2025, LinkedIn began sharing advertising revenue with creators for the first time, specifically tied to video content. The platform also launched a program placing video ads alongside influencer content, explicitly because video drives more time spent on the platform, which is the metric that matters most to LinkedIn’s advertising business.
None of that is sinister. It’s just a normal incentive that doesn’t always line up with what gets your individual post the most reach. LinkedIn benefits when more video gets uploaded and watched in aggregate, because that’s ad inventory and platform stickiness. Whether your specific video gets strong distribution from the ranking algorithm is a separate question with a separate answer, and right now that answer often favours other formats.
Worth knowing: a platform’s promotional messaging about a format and that platform’s actual ranking mechanics aren’t always pointing in the same direction. Test your own results before betting your content strategy on either one.
I’ve seen this exact pattern before
If this feels familiar, it should. Back in 2015 and 2016, the entire media industry went through a “pivot to video,” with publishers gutting their writing teams to chase what Facebook insisted was the future of content.
It turned out Facebook had been reporting video watch-time numbers to advertisers that were significantly inflated, by some estimates 60 to 80%, with later legal filings alleging the real gap was far larger. Facebook eventually settled a related lawsuit with advertisers for $40 million. Many newsrooms had already cut staff and built entire strategies around numbers that didn’t reflect reality.
I’m not saying LinkedIn is doing the same thing. There’s no evidence of that here. What I am saying is that “the platform is heavily promoting this format” and “this format is actually working for individual creators” have been two different questions before, on a platform just as confident as LinkedIn is now. It’s worth treating platform enthusiasm and your own performance data as two separate inputs, not one.
What this means for your content mix
So where does that leave your actual posting strategy? Here’s how I’d think about it:
- Don’t abandon video, but don’t lead with it blindly. If your audience genuinely engages with your video content and your own analytics show strong watch time, keep going. The aggregate data doesn’t override what’s working for your specific account.
- Test document carousels if you haven’t. Given how consistently they’re producing the dwell time signal the algorithm rewards, a carousel built from your best existing insight is a low-cost experiment most accounts haven’t tried seriously.
- Watch your own dwell time and saves, not just views. A video racking up impressions but losing people after three seconds is telling you something the view count won’t.
- Keep links out of the post body either way. Regardless of format, posts with external links continue to see meaningfully reduced reach across every analysis we found. Put the link in the first comment if you need one.
- Re-test quarterly. This is a platform whose ranking model changed meaningfully within the last year. Whatever’s working today is worth checking again in three months.
The takeaway
LinkedIn’s algorithm was never really “pro-video” or “anti-video.” It’s pro-dwell-time, and right now, document carousels are producing that signal more reliably than most video content does. That can change again, and probably will. The move isn’t to chase whichever format a platform is promoting this quarter. It’s to understand the actual mechanism the algorithm is measuring, then test your own content against it directly.
Pull your last ten LinkedIn posts and sort them by save rate and average view duration, not just impressions. That’s the number that’ll actually tell you what to make next.