All In
FR EN
Back to blog
Création de contenu & formats performants 18 min read

LinkedIn Carousels in 2026: The Data Behind the 21% Number

| By Patrick de Carvalho

Contents

A LinkedIn carousel, technically a PDF you upload and readers swipe through, posted a median engagement rate of 21.77% in 2026, according to Buffer's analysis of more than 52 million posts across ten platforms. That is roughly six times what a plain text post gets on the same network. I have been on LinkedIn since April 2004, and I have watched a lot of formats get crowned "the future of the feed" before quietly dying. This one has the data to back the hype, plus a spec sheet that changed twice in the past year and that most of the guides floating around still get wrong. Here is what the numbers show, what the platform actually allows in 2026, and the structure I use to build one in about thirty minutes.

In short: LinkedIn carousels (uploaded PDF documents, swiped page by page) posted a 21.77% median engagement rate in 2026 across 52 million+ posts, per Buffer, roughly six times a text post's 3.18%. Socialinsider's separate 1.3-million-post dataset puts native documents at 7.00% average engagement, still the top format on the platform. The two studies disagree on the size of the win because they measure different things, not because either one is wrong. Current specs: PDF, 1080×1350 px portrait or 1080×1080 px square, under 3 MB, 8 to 12 slides out of a 20-slide maximum.

"Carousel" on LinkedIn means two different things depending on who you ask, and mixing them up wastes time.

LinkedIn shipped a native multi-image carousel feature in 2022. It let you upload several images that swiped like a slideshow. The company killed it on December 14, 2023, along with in-image linking and profile videos, and deleted previously posted carousels from the app, as Social Media Today reported at the time. That feature is gone. It is not coming back for organic posts, though a version of it still exists inside paid carousel ads.

What everyone calls a "carousel" today is something else: a document post. You upload a PDF (or a PPTX or DOCX, which LinkedIn converts), and the platform renders it as a swipeable set of pages, right there in the feed, no click required. It has been available since long before 2023 and it is the only swipeable multi-page format left on organic LinkedIn. When Buffer, Socialinsider, or anyone else publishes 2026 carousel engagement numbers, this is the format they are measuring. The rest of this article uses "carousel" and "document post" as the same thing, because on LinkedIn today, they are.

One property sets it apart from every other format on the platform: it is downloadable. A reader can save your PDF to their device and open it again next week. On a network full of professionals hunting for a template, a checklist, or a framework, that changes what the format is for. It is not just a post. It is a piece of content someone chooses to keep.

What the engagement data actually says

Two separate studies, both published in March 2026, both credible, and they land on different numbers because they counted different things.

Buffer's report, published March 5, 2026, covers more than 52 million posts across ten platforms. On LinkedIn specifically, it breaks down engagement by format like this:

Format Median engagement rate
Carousel (document post) 21.77%
Video 7.35%
Image 6.52%
Link post 3.81%
Text 3.18%

Buffer's own report is honest about the spread inside that carousel number, and this is worth reading before you get excited: "Among weaker-performing carousel posts, it was around 5.4%, which is pretty close to the median rate for video and images. So even a below-average carousel is doing about as well as a typical video or image post." Their strongest carousels cleared 41%. A weak one still lands where a decent video does. The format has a floor, not just a ceiling.

Socialinsider's benchmark, published March 16, 2026 from 1.3 million business posts across 16,645 LinkedIn pages, tells a more modest story: native documents averaged 7.00% engagement in 2026, up 14% from 6.10% in 2025. Still the best-performing format in their dataset, ahead of multi-image posts at 6.45%, video at 6.00%, images at 5.30%, and text at 4.50%. But nowhere close to Buffer's 21.77%.

Why the gap? Different samples (52 million posts across ten platforms versus 1.3 million LinkedIn business posts), different metrics (Buffer reports a median across creator accounts, Socialinsider an average across business pages), and different populations (individual creators tend to post more polished carousels than brand pages do). Neither number is fake. Neither is universal. What both studies agree on, independently, using different data: the carousel is the single best-performing format on LinkedIn in 2026, by a wide margin, regardless of how you slice it.

Why a swipe outweighs a like

LinkedIn's own engineering team explained the mechanism behind this back in 2020, in a post titled "Understanding dwell time to improve LinkedIn feed ranking," written by senior engineer Siddharth Dangi and three colleagues. The core idea: the platform tracks how long a post stays visible on a member's screen, calls it dwell time, and uses a model (they named it P(skip)) to lower the rank of updates people scroll past too fast. In offline testing, adding this signal improved the feed's ranking accuracy by as much as 10%. That model is six years old now, but the principle it introduced, that attention duration outranks a quick reaction, still runs underneath every format decision LinkedIn makes.

A carousel is built to generate exactly that signal. Every swipe keeps the post on screen. A ten-slide document read start to finish holds a reader's attention far longer than the half-second it takes to glance at a photo or skim three lines of text. LinkedIn has never published an official document-specific dwell-time figure, so treat any precise "X seconds" claim you read elsewhere with suspicion, mine included if I ever slip and give you one. What we do have is Dataslayer's own tracked comparison, published in February 2026 and updated in July: across their monitored accounts, document posts pulled 40.5% engagement against 10.7% for other formats combined. Not a controlled study. One vendor's own dataset. Still, it points the same direction as everything else here.

This is the mechanism the All In algorithm framework tracks in more detail: not "does the algorithm like documents," but "which formats naturally produce the signal the ranking model rewards." A carousel does not get privileged treatment because LinkedIn favors PDFs. It gets rewarded because the format makes people stay.

The spec sheet: dimensions, weight, slide count

Half the carousel guides published in the past year cite outdated numbers. Here is what actually holds as of August 2026, per Postiv AI's specification guide, last updated August 3, 2026.

Spec Current value
Portrait dimensions 1080 × 1350 px (4:5)
Square dimensions 1080 × 1080 px (1:1)
Landscape Not recommended; shrinks on mobile
File format PDF preferred (also accepts PPT, PPTX, DOC, DOCX)
Max file size 100 MB (best practice: under 3 MB)
Max slide count 20
Recommended slide count 8 to 12
Minimum body text 24 pt
Minimum heading text 48 pt
Safe zone Central 880 × 880 px, 80 px padding on every side

Two things worth flagging. First, the slide ceiling is 20, not the 300 pages that some 2025-era guides still quote; LinkedIn tightened this at some point without much announcement, so if you are working from an older brief, check it again. Second, portrait beats square for a reason that has nothing to do with aesthetics: roughly three-quarters of the time people spend on LinkedIn happens inside the mobile app rather than a desktop browser, per a 2026 compilation of Buffer and Sprout Social data (the exact split shifts depending on whether native app sessions get counted, but every version of the number points the same way). A taller frame occupies more of that mobile screen without getting cropped, and a post that fills more of the screen is a post that is harder to scroll past by accident.

Keep the file under 3 MB. LinkedIn allows up to 100 MB, but nobody is impressed by a document that takes four seconds to render on a phone with two bars of signal.

A 10-slide structure that holds up

This is the sequence I use, and the one I coach clients through. It is not the only structure that works, but it holds up across enough industries that I stopped tweaking it a while ago. Call it the All In carousel blueprint.

Slide 1, the hook. One job: stop the thumb. A specific number, a contrarian claim, or a promise you can actually keep. No logo, no "hi everyone," no throat-clearing. "3 prospecting mistakes costing you 40% of your leads" works. "5 tips for better LinkedIn content" does not, because it could have been written by anyone, about anything, in any year.

Slide 2, the problem. Two or three lines that name the situation your reader is already living through. Speak to them directly. This slide's only function is to make them think: that's me.

Slides 3 through 8, the substance. Six slides, one idea each. A six-step process, six mistakes with their fixes, six pillars of a strategy, whatever your content actually is. Every slide follows the same shape: a title under ten words, twenty to forty words of explanation, one visual element if it earns its place.

Slide 9, the recap. A numbered list of the six points, one line apiece. Someone who swiped through in eight seconds should be able to screenshot this single slide and walk away with the gist.

Slide 10, the ask. A direct question or a specific invitation. "Which of these are you already doing?" beats "thoughts?" every time, because it asks for something the reader can actually answer in ten seconds.

The rule underneath all ten slides: one idea per slide, never two. A reader swipes through a slide in one or two seconds. If a slide needs more than three seconds to land, it is carrying too much, and it needs to become two slides.

Building one without a designer, in about half an hour

Nobody running a company under 250 people has a designer on call for a weekly post. Fine. You do not need one.

Canva covers most of this. Its free plan is enough to start; the Pro tier, at $18 a month or $144 billed annually, adds brand kit features that keep every carousel you publish visually consistent, which matters more for recognition than any single slide's design. Search "LinkedIn carousel" inside Canva and you will find hundreds of ready templates at the right dimensions.

Google Slides or PowerPoint work as a free, no-frills alternative. Set a slide master at 1080×1350 px, apply your two or three brand colors, export to PDF. It looks less flashy than a Canva template and it takes about the same time to produce.

AI carousel tools (Taplio, Draftly, and a handful of similar products) will generate slides from a prompt or a block of text. I have tried most of them. They save time on the blank-page problem and they are worth using for a first draft. What they do not do is replace a real edit pass: the raw output tends to read like nobody in particular wrote it, which is the opposite of what a carousel is supposed to accomplish. Run it, then rewrite the parts that sound like everyone else's carousel, because on a platform serving hundreds of millions of pages of near-identical AI output, that is precisely what gets scrolled past.

Once you have a template configured, the actual production loop looks like this:

Step Time What happens
Pick the topic 3 min A real problem a client raised recently, or something you got wrong
Outline 10 slide titles 5 min Follow the blueprint above
Write the copy 10 min 20 to 40 words per slide, plain sentences
Design in Canva 8 min Drop into your saved template, adjust text and colors
Export to PDF 1 min Check the file is under 3 MB
Write the LinkedIn caption 3 min Hook, one line of context, a direct CTA

Thirty minutes, once the template exists. Building that first template takes closer to an hour. Every carousel after that is thirty minutes, not thirty minutes plus a design brief.

Five mistakes that quietly kill a carousel's reach

Cramming each slide with text. The instinct to say everything on one slide is the fastest way to kill a carousel. If a slide runs past forty words, split it into two.

Wasting slide 1. It shows up in the feed before anyone taps anything. "Best practices for B2B marketing" creates zero tension. "73% of B2B leaders waste their LinkedIn budget, here's where" forces a decision.

Skipping the caption. The text above the document is not decoration. It shows up in notifications and feed previews, and it gives the algorithm context before a single slide loads. A caption of 150 to 300 words, hook first, direct ask last, measurably extends reach.

Changing the visual style every time. A carousel where each slide uses a different font and color scheme reads as amateur, regardless of how good any individual slide looks. One template, reused, builds a recognizable signature faster than any single great design.

Publishing at random times. LinkedIn weights the engagement your post gets in its first sixty minutes heavily when deciding how far to distribute it. Post when your specific audience is active rather than defaulting to a generic "best time" list; Tuesday through Thursday, late morning to early afternoon in your audience's time zone, is the safest starting point according to the largest 2026 studies, but your own analytics will beat any general rule within a month.

A carousel should not be the only thing you post. It should be the post everything else supports.

For a founder or a sales leader publishing three to four times a week, one carousel functions as the anchor: your best framework, your sharpest retrospective, the thing you actually want your next client or hire to read in full. The other two or three posts stay lighter: a reaction to something in your industry, a short story from the past week, a question to your network. The carousel demonstrates expertise and gets saved. The shorter posts keep you visible between carousels.

A carousel also does not have to be single-use. Each slide can become a standalone text post. The recap slide can turn into a one-image summary for other platforms. The whole thing can be expanded into a blog article or a newsletter section. We track that kind of algorithm behavior in more depth on the All In blog, including how format choice interacts with topical consistency on a profile over time.

What I don't know

I do not know which of the two engagement numbers in this piece, Buffer's 21.77% or Socialinsider's 7.00%, is closer to what your specific account would see if you started posting carousels tomorrow. Both are real, both are dated 2026, and they disagree by a factor of three because they measured different populations. Pick neither as gospel; use them as a floor and a ceiling.

I do not know LinkedIn's exact algorithmic weight for a swipe versus a like versus a save. The company has published the mechanism (dwell time, back in 2020) but not a current, format-specific multiplier, and every "X% boost per swipe" figure circulating on marketing blogs right now traces back to nobody's disclosed methodology. I left those numbers out of this piece on purpose.

I do not know how long the 20-slide cap or the current dimensions stay fixed. LinkedIn adjusted document specs at least once in the past year without a formal announcement. If the numbers in this article are stale by the time you read it, that will be why, and I will update it when I catch it.

FAQ

There is no single number. Buffer's 52-million-post study puts the median at 21.77%, with weaker carousels around 5.4% and top performers above 41%. Socialinsider's 1.3-million-post business-page dataset averages 7.00%. Treat anything in the 5 to 10% range as solid and anything above 15% as strong, and compare your own carousel against your own account's history rather than against either study directly.

Eight to twelve is the range every current source converges on. LinkedIn technically allows up to 20 slides, but completion rates drop noticeably past 12 to 15. Below six slides, the content usually does not justify the format; a text post would do the same job.

Portrait, 1080×1350 pixels (4:5 ratio), is the strongest choice for a mobile-first feed. Square, 1080×1080 pixels, is the fallback for content shared across multiple platforms. Landscape is not recommended; it shrinks on mobile screens. Export as a PDF under 3 MB, even though LinkedIn technically accepts files up to 100 MB.

Can I still post a native multi-image carousel on LinkedIn?

No. LinkedIn removed its native multi-image carousel feature for organic posts on December 14, 2023, and deleted previously published ones. The only swipeable multi-page format left on organic LinkedIn is the uploaded PDF document post, which is what every 2026 carousel study measures. Multi-image carousels remain available only inside paid LinkedIn ad campaigns.

Google Slides or PowerPoint, both free, get you a professional result: set the slide master to 1080×1350 px, apply your brand colors, export to PDF. Canva's free plan works the same way with more templates available. Canva Pro, at $18 a month, adds a brand kit that keeps every carousel visually consistent, which is worth it once you are publishing weekly.

Why do LinkedIn carousels outperform video and text posts?

Carousels keep a reader's attention on screen longer than any other format, and LinkedIn's own engineering team has documented since 2020 that dwell time, the duration a post stays visible before someone scrolls past, directly affects how far a post gets distributed. Each swipe through a carousel extends that duration. No format-specific multiplier has been published, but every available 2026 dataset shows carousels ranked first regardless of methodology.

How often should a B2B account publish carousels?

Once a week works for most founders and sales leaders already posting three to four times weekly. That cadence keeps carousel quality high without turning every post into a production project. Publish a second carousel only when a topic genuinely needs the depth; forcing a simple idea into ten slides usually reads as padding.

Sources

  1. Buffer, State of Social Media Engagement 2026, March 5, 2026: 52M+ posts across 10 platforms; LinkedIn format engagement rates, weaker/stronger carousel range.
  2. Socialinsider, LinkedIn Organic Benchmarks 2026, March 16, 2026: 1.3M business posts, 16,645 pages; native document engagement 7.00%, year-over-year change, format comparison.
  3. LinkedIn Engineering, Siddharth Dangi et al., "Understanding dwell time to improve LinkedIn feed ranking," May 12, 2020: official mechanism, P(skip) model, ranking accuracy improvement.
  4. Dataslayer, "LinkedIn Algorithm 2026: What Works Now," February 12, 2026, updated July 2026: document post engagement test, 40.5% versus 10.7%.
  5. Social Media Today, "LinkedIn Says Past Carousel Posts Will Be Deleted Next Month," 2023: removal of native multi-image carousels, December 14, 2023.
  6. Postiv AI, "LinkedIn Carousel Dimensions 2026: Size, Specs & Format Guide," updated August 3, 2026: current dimensions, file size, slide cap, typography minimums, safe zone.
  7. SQ Magazine, "LinkedIn Mobile vs. Desktop Statistics 2026": mobile time-spent compilation from Buffer and Sprout Social data, methodology caveat.
  8. SocialRails, "Canva Pricing 2026," 2026: Canva Pro pricing, $18/month, $144/year.

All In: the format that rewards a sharp idea over a loud one

A carousel does not win because it looks good. It wins because it holds attention long enough for LinkedIn's own ranking signals to notice, which is the same principle behind everything we cover on this blog.

All In is the B2B media that decodes LinkedIn, expert blog, weekly podcast and newsletter for SME leaders and sales directors who want to turn LinkedIn into measurable growth. An original creation by Patrick de Carvalho, on LinkedIn since 2004. Motto: "I Never Lose."

Discover All In and get what matters on LinkedIn, every week.