In the published 2026 data, carousels (PDF document posts) usually win on engagement rate and saves. Native video is where LinkedIn itself is putting its effort, and it can drive strong lead conversion when the video is short and captioned. Neither format wins on every metric. Pick carousels for frameworks and reference material, pick video for trust and personality, and test both on your own account before you believe anyone's benchmark, including mine.
Key Takeaways
- Most third-party benchmarks show carousels with higher engagement rates than video, but samples and methods differ a lot. Treat any single number carefully.
- LinkedIn has been pushing video hard. The company told Digiday that total video viewership rose 36% year over year in late 2024 into early 2025.
- In one creator's 500-post test, carousels got the higher engagement rate (7.8% vs 5.9%) and video got the higher lead conversion (4.7% vs 3.4%).
- Carousels suit frameworks, checklists and data. Video suits stories, opinions and showing your face to build trust.
- Run a 6-8 week split test on your own audience. Track saves, ICP comments and DMs, not just impressions.
Why This Comparison Is Harder Than It Looks
Search "LinkedIn carousel vs video 2026" and you'll find confident headlines and precise-looking percentages. Before you build a strategy on them, keep in mind:
- Most benchmarks come from tool vendors or individual creators. Their samples lean towards their own users and niches.
- "Engagement rate" isn't defined the same way everywhere. Some divide by impressions, some by followers, and some count carousel slide clicks as engagement, which inflates carousel numbers.
- Format and content get mixed up. A great carousel versus a lazy video tells you about quality, not format.
- The algorithm changes. A benchmark from early 2025 may not describe September 2026.
So I'll show you the data, say where it comes from, and then give you a way to get your own answer.
What LinkedIn Has Said About Video
The most reliable video data comes from LinkedIn itself, even if it describes platform growth, not per-post performance. In January 2025, a LinkedIn representative told Digiday that:
- Total video viewership grew 36% year over year (late October 2024 to late January 2025).
- Video uploads grew 34% year over year in its fiscal Q4 2024.
- Video creation was growing at twice the rate of other post formats.
LinkedIn has also added video-first surfaces to the app, like the vertical video feed. When a platform invests like that, it usually wants more supply of that format. That's a reason to take video seriously, but it doesn't guarantee your videos will reach more people than your carousels.
Third-Party Benchmarks: Carousel vs Video
Here's one of the more transparent creator tests. Baz Furby of Grow with Ghost reports results from 500+ of his own posts between January and November 2025:
| Format | Avg. impressions | Engagement rate | Lead conversion |
|---|---|---|---|
| Text | 2,847 | 4.2% | 2.1% |
| Carousel | 1,924 | 7.8% | 3.4% |
| Video | 1,543 | 5.9% | 4.7% |
| Document (short template) | 1,234 | 6.1% | 5.8% |
| Poll | 2,156 | 8.4% | 1.2% |
Source: Grow with Ghost, one account's data, reported December 2025. Not a platform-wide benchmark.
What I take from this:
- Carousels beat video on engagement rate in this dataset, which matches the direction of most vendor benchmarks I've seen.
- Video converted better to leads. That fits what I see with clients. People who watch you talk for 60 seconds feel they know you.
- Text still got the most reach. Format isn't everything.
- Polls get engagement but little value. The author notes most poll engagement is votes, not conversation.
Other roundups you'll find online cite carousel engagement rates in the 6-7% range and claim carousels outperform other formats by large multiples. Many don't show their methodology. I'd treat those as directional at best.
When Carousels Win
From my own client work and the data above, carousels are the better pick when:
- The content is reference material. Frameworks, checklists, step-by-step processes, "7 mistakes" breakdowns. People save these to come back to.
- You have data to show. Charts and before/after comparisons read better on slides than in a talking-head video.
- Your audience reads on desktop at work. Swiping a PDF at a desk is easy. Watching video with sound in an open office isn't.
- The creator isn't comfortable on camera. A good carousel beats a nervous video every time.
Carousel best practices
- Slide 1 is the hook. A specific promise ("How we cut cost per lead 38% in 6 weeks"), not a vague topic.
- One idea per slide. Large text, readable on a phone.
- 8-12 slides. Enough to deliver value, short enough to finish. Many creators recommend 9-10.
- End with a summary slide, not only "Follow for more." Summary slides get screenshotted and shared.
- Put the key point in the post text too. Search engines and AI systems read post text far more reliably than slide images.
When Video Wins
Video is the better pick when:
- Trust is the goal. Founders, consultants and coaches selling high-consideration services gain a lot from being seen and heard.
- The content is a story or an opinion. Tone, pauses and conviction come through on video in a way slides can't match.
- You're demoing something. Product walkthroughs, screen recordings, before/after live edits.
- You want reuse across platforms. One vertical video can go to LinkedIn, Instagram Reels and YouTube Shorts with small edits.
Video best practices
- Hook in the first 3 seconds, both spoken and as on-screen text.
- Captions are mandatory. Many people watch with the sound off.
- Aim for 30-90 seconds for feed video. The Grow with Ghost data put its best results at 45-90 seconds.
- Vertical or square framing for mobile.
- Write a real post caption. Summarise the main point in text so the post works for people who don't press play, and for AI systems that can't watch it.
The Hybrid Approach I Recommend
For most B2B accounts I work with, the answer isn't either-or. It's a rotation built around one idea a week:
| Day | Format | Same core idea, different job |
|---|---|---|
| Tuesday | Carousel | The framework: steps, checklist, data |
| Thursday | Short video (45-75s) | The story: why it matters, what went wrong, your opinion |
| Saturday | Text post | The sharpest single insight, written to be quoted |
This squeezes more out of each idea, lets the algorithm test you in several formats, and gives you clean comparison data in a few weeks.
When I worked on organic social for Masai School during its LinkedIn growth from 50K to 160K followers, format variety mattered. What mattered more was that every format carried something real, like a student outcome, a number or a strong opinion. Format helps good content travel. It won't save weak content.
How to Run Your Own Format Test
Benchmarks describe other people's audiences. Here's the 8-week test I run with clients.
Setup
- Pick 8 core ideas of similar strength.
- Make each idea as both a carousel and a video. Randomise which format goes out first.
- Keep variables steady: same posting window, similar text length, no engagement pods.
What to track (per post)
- Impressions and members reached
- Engagement rate (reactions + comments + reposts ÷ impressions; define it once and stick with it)
- Comments from your ideal customer profile (tag them by hand)
- Saves and sends, where your analytics show them
- Profile views in the 48 hours after posting
- DMs or inbound leads that mention the post
How to decide
After 8 weeks, compare the median (not the average, since one viral post skews everything) on your primary goal. If the goal is pipeline, weight ICP comments and DMs most. If it's audience growth, weight reach and follows.
The AI-Visibility Angle
There's a newer factor in format choice: AI answer engines now cite LinkedIn heavily. A Semrush study of 89,000 LinkedIn URLs cited in AI search found long-form articles made up the majority of citations, and cited feed posts were typically short (50-299 words).
Neither carousel slides nor video audio are reliably readable by AI systems. So if you want a post to be citable, the post text has to carry the core claim, whatever the format. I cover this in detail in my post on turning LinkedIn posts into AI-citable assets.
Common Mistakes
- Uploading landscape slide decks as carousels. Tiny text on mobile kills completion.
- Posting YouTube links instead of native video. LinkedIn tends to favour native uploads over outbound links.
- Judging by one post. One format test is noise. Eight is a signal.
- Chasing engagement rate alone. Polls "win" on engagement rate and often lose on business value.
- No captions on video. It's the single most common and most avoidable mistake.
FAQ
Are LinkedIn carousels still effective in 2026?
Yes. Most available benchmarks show carousels with some of the highest engagement rates on the platform, especially for frameworks and data. The numbers vary by source and methodology, so test on your own account.
Is LinkedIn prioritising video in 2026?
LinkedIn has invested heavily in video, including a vertical video feed, and has reported strong growth in video viewing and uploads. That suggests video is strategically important to the platform. Individual post reach still depends on content quality and audience response.
Which gets more reach on LinkedIn, carousel or video?
It depends on the account and the content. In one creator's 500-post dataset, carousels averaged more impressions than video, but text posts beat both. Run your own split test for a reliable answer.
Which format generates more leads on LinkedIn?
In the Grow with Ghost dataset, video (4.7%) converted better than carousels (3.4%), and short template documents converted best of all. That's one account's data, but it matches a common pattern: formats that build personal trust, or give away something useful, tend to drive leads.
How long should a LinkedIn video be?
For feed video, 30-90 seconds is a safe range for most B2B content. Put the hook in the first three seconds and add captions. Longer videos can work for webinars or deep dives, but completion drops.
How many slides should a LinkedIn carousel have?
Most practitioners recommend 8-12 slides, with many landing on 9-10. Use one idea per slide, large readable text, and finish with a summary slide.
Can I turn one piece of content into both a carousel and a video?
Yes, and I recommend it. Use the carousel for structure (steps, frameworks, data) and the video for the story and your opinion on the same idea. It doubles your output per idea and gives you format comparison data.
Do AI tools like ChatGPT read LinkedIn carousels and videos?
Not reliably. AI systems mainly rely on the text of a post or article. Always summarise the main claim in the post text so it can be understood and cited without the slides or audio.
Let's Find the Format Mix That Works for You
I've spent 4+ years in marketing testing formats across LinkedIn, Instagram and more for startups and edtech brands, and the right answer has always come from the account's own data. If you want help designing a format test or a content system that fits your team, look through my work and reach out via the contact form at younusfardeen.in.