LinkedIn carousels in 2026: how to make document posts that get saved

A LinkedIn carousel is a PDF uploaded as a document post: readers swipe through it slide by slide, inside the feed, without leaving LinkedIn. Carousels earn two things the algorithm cares about, dwell time (people stay to swipe) and saves (people come back later), which is why they consistently outperform text posts on those two signals. They are also the format most people get wrong, usually by cramming a blog post into ten slides.

This guide covers when a carousel is the right format, how to structure one, how to design it for a phone, the technical specs LinkedIn expects, and how to make them without a designer.

What counts as a carousel on LinkedIn

LinkedIn does not have a native "carousel" post type any more; the swipeable format is a document post, a PDF, PowerPoint or Word file uploaded through the document option. LinkedIn renders each page as a slide. Multi-image posts exist too, but they display as a grid rather than a swipe, and they do not earn the same dwell time. When people say "LinkedIn carousel" they mean a PDF document post.

When a carousel is the right format

Format should follow the idea, not the other way round. A carousel earns its place when the content is a sequence:

• A process with steps (3 to 8).
• A list where each item deserves its own beat ("5 mistakes", "7 questions to ask").
• A comparison: before/after, myth/reality, us/them.
• A framework you can name and walk through.
• A teardown: one example, annotated slide by slide.

If the idea is a single story, write a text post. If it needs one chart, use an image. If it is a sequence, carousel.

The structure that works

Slide 1, the cover

One line, large. This is the hook: it has to make the swipe worth it. Treat it exactly like a post hook, a specific promise, a number, a contrarian line, and add a small "swipe →" affordance so first-time readers know it moves. Never put a paragraph on the cover.

Slides 2 to N, one idea per slide

A heading plus one to three short lines. If a slide needs more text, split it into two slides. Each slide should be understandable on its own, because readers drop in and out; but the sequence should build, because readers who finish are the ones who save.

The last slide, the close

A one-sentence summary, a question, or a clear "what to do next". This is where saves and comments happen. A CTA to follow you is fine; a CTA to buy something is usually not, unless the carousel was explicitly a product walkthrough.

How many slides

Aim for 6 to 9. Under five feels thin and does not earn dwell time; over twelve loses people before the close. If you have fifteen points, you have two carousels.

Design rules for a phone

Carousels are read at phone size, in a feed, often with the sound off and in a hurry.

• Portrait or square. 4:5 (1080×1350) is the sweet spot: it fills the mobile screen. Square (1080×1080) is fine. Landscape wastes half the screen.
• One visual system. Same background, same typeface, same accent colour, same margins on every slide. Consistency is what makes a deck look designed rather than assembled.
• Big type. Headings 60–80 px at 1080 wide, body no smaller than 36 px. If you have to zoom, it is too small.
• High contrast. Dark background with light type, or the reverse. Avoid mid-tones on mid-tones.
• A page marker. "3/8" in a corner tells readers how long the ride is and nudges them to finish.
• Generous whitespace. One idea, lots of air. Crowded slides get skipped.
• No tiny screenshots. If you must show a UI, crop to the part that matters and enlarge it.

Technical specs

• File type: PDF (most reliable), PPT/PPTX or DOC/DOCX also accepted.
• Size limit: 100 MB and 300 pages, you will never get close; keep it under 5 MB for fast loading.
• Dimensions: any, but 1080×1350 or 1080×1080 render best on mobile.
• Fonts: embed them or export as images-in-PDF, so the rendering does not depend on the reader's device.
• Title: LinkedIn asks for a document title on upload; it is shown above the carousel and used for accessibility, make it descriptive, not "Untitled".

The caption still matters

The document is the content; the post text above it is the hook and the context. Two or three short lines: what the carousel is, who it is for, why now. Then let the carousel do the work. Do not repeat the slides in the caption, it kills the reason to swipe.

Mistakes that make carousels flop

• The blog post in disguise. Ten slides of paragraphs. Nobody swipes through prose.
• The weak cover. "Some thoughts on hiring" is not a reason to swipe. "The interview question that predicts who quits in 90 days" is.
• Inconsistent slides. Different fonts and colours from slide to slide reads as sloppy, and sloppy does not get saved.
• No close. The deck just ends on point seven. Give the reader something to do.
• Too many words per slide. If the reader has to stop scrolling to read, they scroll on instead.
• Publishing at the wrong moment. Carousels earn dwell time, so give them a good first hour, see the best time to post.

Making them without a designer

The bottleneck used to be production: designing nine coherent slides for every list post is an afternoon in a design tool. Two ways to shortcut it:

• A template. Build one slide system, cover, content slide, close, and reuse it. Consistency is free once the template exists.
• Generation. In ShipMyPosts, any post can become a carousel in one click: the content is planned slide by slide, each slide is generated as an image that carries the style of the previous ones, and the deck is assembled into the PDF LinkedIn expects. You review it on the calendar like any other post, and it publishes as a document post on schedule.

Either way, the rule is the same: the deck is only as good as the sequence behind it. Plan the slides as sentences first; design second.

Five carousel ideas you can make this week

If you are not sure what deserves the format, start with one of these. Each is a sequence by nature, which is the whole test.

• "The N questions I ask before [decision]." Cover: the decision. One question per slide with a one-line reason. Close: the question people forget.
• "Before / after: how we [changed a process]." Alternate slides: before, after, before, after. Close: the metric that moved.
• "[Myth] vs what actually happens." Three to four pairs. Close: the one myth that costs the most.
• "How we [did a thing] in N steps." The classic. Keep each step to a heading and one line; put the detail in the caption's first comment.
• "A teardown of [one example]." One screenshot per slide, cropped and enlarged, with a one-line annotation. Close: the pattern across all of them.

Turning a text post into a carousel

Your best-performing text posts are carousel material. Take the post, break it into its beats, usually the hook, four to six points, and the ending, and give each beat a slide. The hook becomes the cover verbatim. The ending becomes the close. What you cut is the connective tissue; what you keep is the spine. Republish the carousel two to three weeks after the text post: most of your audience did not see the original, and the ones who did will save the version they can flip through.

Measuring a carousel

Carousels earn different signals from text posts, so judge them differently:

• Saves: the primary signal. A carousel with more saves than reactions did its job.
• Dwell time: visible indirectly as impressions that keep growing on day two.
• Comments: usually fewer than a strong text post; that is normal.
• Profile views: carousels are a strong "who is this?" driver because they look like effort.

If a carousel gets reactions but no saves, the content was not worth keeping; if it gets saves but no reach, the cover failed. Fix the cover first, it is the hook, and the hook decides everything.

A carousel is a sequence people want to finish. Everything else is decoration.