LinkedIn carousel posts: the honest data, and how to build one

LinkedIn carousel posts are document posts: you upload a PDF, PowerPoint, or Word file and the feed turns it into swipeable slides. In our analysis of 12,988 LinkedIn posts, document posts made up 1.2% of the top-performing decile and 0.1% of the bottom half, a 12x over-index and the widest format gap in our data. Now the part most articles skip: that gap rests on 60 carousel posts from 13 accounts, 15 of them in the top decile. The sample is too small to conclude anything. Treat it as a hint worth testing on your own account, not as proof.

What is a LinkedIn carousel post?

There is no button in the LinkedIn composer labelled “carousel.” What people mean by a LinkedIn carousel post is a document post. You click the document icon, upload a file, give it a title, and LinkedIn converts each page into a slide that readers swipe or click through inside the feed. A page counter sits in the corner. The whole thing plays without leaving the feed, which is the entire point.

Two things get confused with this. The first is the carousel ad, a paid format made of two to ten square images with individual destination links. That is a LinkedIn Ads product, it costs money, and it behaves nothing like an organic document post. The second is the multi-image post, where you attach several photos to a normal update. Multi-image posts are useful, but they are not documents: no page counter, no PDF, no slide-by-slide pacing.

The organic document post has one property that no other format on LinkedIn has: it holds attention inside a fixed frame. A text post is read or skipped in one motion. A document post asks for a swipe, then another, then another. Each swipe is a small act of engagement, and the time spent doing it is time the post stays on screen. Whether LinkedIn weights that directly is not something we can measure from outside, but the mechanic is obviously different from every other format.

Documents also survive the trip out of the feed. People save them, screenshot them, and send them to colleagues. A checklist someone downloads on Tuesday can bring you a reply on Friday. That is a slower loop than a viral text post, and it is a more durable one.

What did 12,988 posts say about carousel performance?

We scored 12,988 English LinkedIn posts by engagement rate and compared the top decile (1,298 posts) against the bottom half (6,494 posts). Here is where every format landed, with the raw cohort count for each so you can see what each percentage is actually standing on.

The 12x ratio is the biggest number in our whole format analysis, and it is the one we trust least. Compare it with the video row directly underneath: video's 2.6x over-index is a smaller multiple built on 1,613 posts from 46 different accounts. That is a finding. The carousel row is 24 posts total across both groups. Move three posts between buckets and the multiple changes shape.

The 60 carousel posts in our cohort also come from just 13 accounts, and roughly a quarter of them come from a single creator who happened to make documents a habit. So we are not really measuring “how carousels perform.” We are measuring how a handful of people who liked making carousels performed. Those are different questions.

That post is worth pausing on, because it shows the mechanic that makes documents work. The value was not the design. It was that the content only existed because the audience made it, so the audience came back to see it. A carousel is a container. What you put in it still decides everything.

Why is 60 posts not enough to call carousels a winner?

Because of how small differences behave in small samples. With 60 observations, a handful of unusually good posts can swing the ratio by several multiples. There is no confidence interval we could publish here that would be narrow enough to be useful. The honest statement is: documents appear more often among our best posts than our worst, and we cannot rule out that this is noise.

There is a second problem, and it is the more interesting one: selection. Nobody makes a carousel by accident. Making one means opening a design tool, writing slide copy, exporting a PDF, and proofreading it. That filter removes almost every low-effort post from the format before we ever measure it. Text posts include everything from a carefully built argument to a one-line congratulations. Carousels include almost nothing casual.

So part of what we are seeing is a proxy for effort, not a property of the format. This is the same trap that makes people over-read the image row in the table above. Images show up more often in the bottom half, which sounds like an argument against images. The likelier reading is that attaching a stock photo is the default dressing on a low-effort post, so the image bucket collects a lot of filler. We wrote about the same confound in the context of link posts and reach, where effort and format are similarly tangled.

And one more caveat specific to this dataset: it captures a slice of LinkedIn from a period when document posts were a newer, rarer format. Formats that are rare tend to get an attention bonus simply by breaking the pattern of the feed. That bonus decays as the format becomes common. If your feed is now full of carousels, the arbitrage that these 60 posts might be capturing is already smaller.

What do larger studies say about document posts?

Our sample is thin, so it is worth checking against people with more data. Socialinsider's LinkedIn benchmarks, built from 1.3 million posts across 16,645 business pages between January 2024 and December 2025, put native documents at the top of their format table with a 7.00% engagement rate, ahead of multi-image (6.45%), video (6.00%), image (5.30%), text (4.50%), polls (4.20%), and link posts (3.25%).

Two caveats before you take that as confirmation. Their sample is company pages, not personal profiles, and page audiences behave differently from personal followings. And their engagement rate is computed per impression, which is a different denominator from ours (we divide by follower count). Different populations, different math. What is useful is the ordering: two independent samples, measured differently, both put documents near the top and link posts at the bottom.

That is about as much as the public evidence supports. Anyone telling you carousels get a specific multiple of normal reach is quoting a number that either has no methodology attached or comes from a sample as thin as ours. The direction is reasonably well supported. The magnitude is not.

How do you make a LinkedIn carousel post, step by step?

1. Write the slides as text first. In a plain document, one line per slide. If the sequence does not make sense as bare sentences, no design will save it. This step takes ten minutes and prevents most bad carousels.
2. Design at square or wide. Square (1080 x 1080) uses the most vertical space in the mobile feed. Wide pages (A4 or US letter, turned on their side) are what LinkedIn recommends and work well for anything with a chart or a table.
3. Export as PDF. LinkedIn accepts PPT, PPTX, DOC, DOCX, and PDF, up to 300 pages and 100MB, per Hootsuite's guide to LinkedIn carousels. Only PDF renders exactly as you designed it. The others get converted, and conversion moves things.
4. Upload via the document icon in the composer, not the image icon. If you attach a PDF the wrong way it becomes a file attachment instead of a swipeable document.
5. Give it a document title.LinkedIn shows this under the slides. It is a real piece of copy, not a filename. “Q3-deck-v4-FINAL” is a wasted line.
6. Write the caption last. The caption is the hook that decides whether anyone swipes at all. More on that below.
7. Check it on a phone before you post. Every time. Text that looks balanced on a laptop is often unreadable at feed size.

You cannot swap the file after publishing. Caption edits are allowed; document edits are not. A typo on slide four means deleting the post and starting over with zero engagement, which is a genuinely bad trade if the post was doing well.

How many slides should a LinkedIn carousel have?

Six to twelve. The ceiling of 300 pages is a technical limit, not a target. What actually matters is the share of readers who reach the last slide, and that number falls with every additional swipe you ask for.

Think of the slide budget in three parts:

• Slide 1: the cover. This is your real hook. It has to work as a still image in a scrolling feed, with no audio and no context. One claim, large type, nothing else.
• Slides 2 to n-1: one idea each. If a slide needs two ideas, it needs to be two slides. If an idea needs two slides, it is probably too big for this format.
• Final slide: the close. One ask. Follow, comment, or save. Not three asks. We cover why a single, specific close outperforms a menu in the guide on how to end a LinkedIn post.

Under six slides and the swipe rarely justifies itself; you had a text post and you added friction. Past twelve, you are asking a distracted reader on a phone for a sustained commitment they did not agree to. If your content genuinely needs twenty slides, it wants to be a LinkedIn newsletter instead, where length is expected and subscribers opted in.

How should each slide be designed?

The constraint that governs everything: your carousel will mostly be seen at about the size of a playing card, held at arm's length, on a phone, by someone doing something else. Design for that and the rest follows.

• Big type, few words. Roughly fifteen to thirty words per slide. If you are shrinking the font to fit, cut the copy instead.
• High contrast. Dark text on light, or light on dark. Mid-grey on off-white loses at thumbnail size and fails accessibility checks besides.
• One visual idea per slide. A number, a short line, maybe one simple graphic. Slides that look like PowerPoint bullets from a 2011 sales deck get swiped past.
• Consistent grid. Same margins, same type sizes, same position for headings on every slide. Consistency is what makes a homemade carousel look deliberate.
• Real alt text in the caption. Document posts are images to a screen reader. If the substance lives only in the slides, summarise it in the caption text so the post is not a blank rectangle for anyone using assistive tech.
• Number the slides.“3/9” in the corner sets an expectation of length, and expectations are what get people to the end.

Skip the branded template on personal posts. A logo on every slide reads as a corporate deck escaped into the feed. What builds recognition is a consistent typeface and colour, which you can keep without stamping your brand on every frame.

When do LinkedIn carousel posts beat a plain text post?

The honest default on LinkedIn is still text. Text posts are cheap to make, cheap to abandon, and in our cohort they held up fine (19.3% of the top decile against 16.6% of the bottom half). A carousel costs an hour or more. It has to earn that hour.

A simple test: if you can say it in 200 words of plain text without losing anything, say it in 200 words of plain text. Carousels earn their keep when the content has structure that text has to describe and slides can just show.

When is a carousel the wrong call?

Three failure modes come up repeatedly, and all three are avoidable.

The padded carousel.One idea stretched across eleven slides with transitions like “But here is the thing.” Readers notice by slide three and leave. The tell is that you can delete any middle slide without losing information.

The repurposed deck. Uploading a client presentation or a webinar deck because the file already exists. Those decks were built for a speaker who explains them. Without the speaker they are a stack of headings.

The carousel used to dodge writing. If the caption is one vague line and the slides are the whole post, you have moved your writing problem into a design tool. The caption is still the hook. Nothing about a carousel changes that the first two lines decide whether anybody sees the rest, which is the argument we make at length in the guide to LinkedIn hooks.

There is also a cadence question. Carousels take real time, and consistency beats production value on LinkedIn. If making carousels means you post twice a month instead of four times a week, the format is costing you more than it returns. Sort out how often you can realistically post before deciding how elaborate each post should be.

What should the caption above the carousel say?

The caption does one job: make the swipe feel worth it. It is not a summary of the slides, and it is definitely not “New carousel, let me know what you think.”

The pattern that holds up:

1. Line one states the payoff. What the reader will be able to do after swiping. Concrete beats clever.
2. Line two gives a reason to trust it. Where the content came from: your own numbers, twelve client projects, a mistake you made twice.
3. Line three points at the document.Short. “Nine slides below.”

Two structural notes from our hook data, which covers all 12,988 posts and is therefore much sturdier than the carousel numbers. Saying you in the visible hook showed up in 47.5% of top-decile posts against 32.7% of the bottom half. And hashtag stuffing runs the other way: posts with four or more hashtags were 41.9% of the bottom half against 25.9% of the top decile, which we broke down in the hashtag analysis. One to three relevant tags, at the end, is plenty.

Keep the caption above the “see more” fold where you can. On mobile that cutoff is tighter than the roughly 210 characters you get on desktop, and everything past it costs a click. Our guide to opening a post covers how to compress an opener without making it sound like a headline.

How do you tell whether your carousels actually worked?

Impressions are the wrong scoreboard for this format. A carousel that reaches 4,000 people and is saved by 40 of them is doing more for you than one that reaches 12,000 and is saved by 3. Watch these instead:

• Saves. The clearest signal that the document had reference value. Saves also mean the post can resurface for that person weeks later.
• Comments per impression. Comments are the scarcer currency. In our cohort the top-decile median was 72 comments against 5 in the bottom half, a far wider gap than reactions (235 against 80).
• Profile views in the 48 hours after. The carousel did its job if people went looking for who made it.
• Reshares. Documents travel further than text because they carry their own context. A reshare of a carousel is worth more than a reshare of a text post.

If you are unsure what any of these numbers should look like at your follower count, start with our engagement benchmarks from 12,988 posts: the median post in our cohort earned 0.40 engagement per 1,000 followers, the 75th percentile 1.27, and the top decile threshold 5.95. Those are the percentile lines to measure yourself against, and the difference between impressions and reach matters before you compare anything.

How do you test carousels on your own account?

Our sample cannot tell you whether carousels work for you. Your own can, and it takes about six weeks. Here is the test we would run.

1. Baseline first. Pull your last 30 posts. Record impressions, reactions, comments, and saves for each. Compute the median, not the average, because one viral post will distort the average badly.
2. Make six carousels over six weeks. One per week, same posting time, same topic area as your usual content. Six is not statistically impressive, but it is six times what most people try before forming an opinion.
3. Hold everything else constant. Same cadence, same hook style, same hashtag habits. If you change three things at once you will learn nothing.
4. Compare medians against your baseline. Then read the comments. If carousel comments are more substantive than your usual ones, the format is reaching a more engaged slice of your audience, and that matters more than the impression count.
5. Decide on effort, not just outcome. If carousels beat your baseline by 20% and cost four times the time, they are a bad trade unless the audience they reach is the audience you want.

Run the same test on native video, where the underlying evidence is considerably stronger, and you will have a real format strategy instead of a borrowed one. Two tested formats and a baseline beat any amount of advice about what works in general.

So should you make LinkedIn carousel posts?

Yes, occasionally, for the right content. LinkedIn carousel posts over-index 12x among our top-decile posts, and a much larger external benchmark independently ranks documents first among formats. That is enough to justify trying them. It is not enough to justify rebuilding your content plan around them, and anyone quoting you a precise reach multiplier for carousels is working from a sample no better than our 60 posts.

The practical version: keep writing text posts, because they are cheap and consistent. Add a carousel when you have something sequential worth six to twelve slides. Put the real work into slide one and the caption. Check it on a phone. Then measure saves and comments rather than impressions, and let your own numbers settle the question that our 60 posts cannot.