Turning One Piece of Content Into a LinkedIn Carousel

You already have the material. A blog post you spent two days on, a talk you gave to forty people, a report nobody outside the company opened, a long LinkedIn post from eight months ago that did unusually well. Repurposing it into a carousel sounds like the cheapest content you will ever make.

It usually is not, because most people do the obvious thing: they compress the source into slides. Ten sections become ten slides, each one a shrunken paragraph. The result reads like a contents page. This article is about the alternative — treating the source as raw material rather than as a script.

First, what exactly am I making?

On LinkedIn today, “carousel” means one of two organic formats, and it helps to know which you are aiming at before you start cutting.

• A document post — you upload a PDF (or a PowerPoint or Word file) and LinkedIn renders it as swipeable pages inside the feed. This is what most people mean when they say “LinkedIn carousel”.
• A multi-image post — several images in one post that the reader swipes through. Fewer slides, no file to build, but no page-like flow either.

Both are free organic posts. Neither is a “carousel ad”, which is a separate paid LinkedIn Ads format and not what this workflow is for. For repurposing longer written material, the document post is usually the better fit because it comfortably holds more slides.

Why does summarising the whole source not work?

A blog post and a carousel are read in completely different conditions. The blog post was chosen — somebody clicked a headline and set aside a few minutes. The carousel is not chosen. It appears between a job announcement and someone’s holiday photo, and it has to earn the second slide from a standing start.

A summary of an 1,800-word article gives every idea equal, shallow treatment. Nothing is developed enough to be useful, and nothing is surprising enough to be worth swiping. You end up with the least interesting version of a piece that was interesting.

The fix is a change of goal. You are not moving the article into a new format. You are extracting one argument from it and building a small, complete thing around that argument. The article becomes a quarry, not a blueprint.

How do I choose the one argument?

Read the source and look for the sentence you would defend in a meeting. Not the topic — the claim. “Onboarding matters” is a topic. “Most onboarding fails in week three, not week one” is a claim. A claim can be agreed with or argued against, which is exactly what makes it worth putting in front of people.

Three practical tests when you have candidates:

• Can you state it in one line, without the source? If it needs three sentences of setup to make sense, it is not ready to be slide one.
• Would a reasonable person in your field disagree, or at least raise an eyebrow? Claims nobody would contest are true but inert.
• Do you have three to six things to say in support? Fewer, and it is a single text post rather than a carousel. Many more, and you are back to summarising.

Long sources often contain three or four workable claims. That is a good problem — it means three or four carousels, published weeks apart, rather than one that tries to hold all of them. Only one goes in this carousel.

How do I map the source onto slides?

Once the claim is fixed, the mapping is close to mechanical. Work from a structure rather than free-styling — a carousel is a sequence, and a sequence with a shape is far easier to write.

Two rules keep this honest. One idea per slide. The moment a slide carries two points, split it or drop one. Each slide should make sense to someone who joined at that slide, because people do land mid-swipe.

Write the whole set as plain text first — ten lines in a document, no design at all. If those ten lines do not hold together as an argument when read straight through, no template will rescue them. This is also the point to check the flow: does slide four follow from slide three, or did you simply keep the order they appeared in the article?

What should I cut?

Cutting is most of the work. Assume the large majority of the source does not survive, and cut in this order:

• The warm-up. Every article opens by establishing that the topic exists. A carousel cannot afford that. Delete the first two paragraphs of the source and start at the first real statement.
• Anything supporting a different claim. This is the hardest cut, because those sections are often good. They are next month’s carousel.
• Caveats and qualifiers. Written prose can carry “in most cases, though it depends on team size”. A slide cannot. Either the point survives without the hedge, or it was too fragile for this format.
• Nested detail. Sub-steps, configuration notes, edge cases. These are why the original article still exists, and why linking to it is worth doing.
• Transitions. “Now that we have covered X, let us turn to Y” does no work at all when the reader is swiping.
• Numbers you cannot source. If a figure in the original was a rough estimate that hardened into a fact over time, take it out. A number on a slide gets screenshotted and repeated, so it needs to be one you can stand behind and attribute.

Keep the specifics, though. Vague slides are the most common failure of repurposed content, and they happen because generalising is the easiest way to shorten something. “Improve your handover process” fits on a slide and says nothing. “Write the handover before the last week, not during it” fits just as well and is worth reading.

Does this work for a talk, a report or an old post?

The claim-and-cut method is the same. What changes is where you dig.

A talk or webinar

Skip your own slides — they were designed to support you speaking, so they are usually too sparse to stand alone. Go to the transcript, and specifically to the questions at the end. The question someone actually asked makes a better slide one than anything in your prepared deck, because it is a real person’s real confusion. Also look for the moment you went off-script to explain something a second way; that improvised version is usually your clearest.

A report or a data-heavy document

Resist rebuilding the charts. One finding, one carousel. Slide one states the finding in words, not as a graph. The middle slides explain what it means for the reader’s own work, and one slide — two at most — carries a simplified chart. Name the source and the sample size on the slide itself, because a chart travels away from its context immediately.

A long post that already did well

This is the easiest source, because the audience already told you which part worked. Read the comments. Whatever people quoted back, argued with, or asked to hear more about is your claim. The carousel is then a deeper treatment of that one thread — not a repeat of the post in bigger type.

Do I need to write a new caption?

Yes. Reusing the source’s introduction as the caption is the most common shortcut, and it reads exactly like what it is. The caption has its own job: give someone who will never swipe a reason to care, and give someone who might a reason to start. Two or three short lines, ending on the question the carousel answers, is plenty. If you want to point back to the original article, many creators put that link in the first comment rather than in the caption — treat that as a long-standing creator habit rather than a documented LinkedIn rule, and test it on your own account.

How do I actually publish it?

Export your slides as a PDF and upload it as a document post, or export them as individual images for a multi-image post. LinkedIn’s own help pages are the right place to check current file size limits, accepted formats and slide maximums, since those details change and are not worth guessing at.

If you are drafting in an AI assistant already, you can publish from there without a second stop. Kalovio connects Claude or ChatGPT to your own LinkedIn account through LinkedIn’s official sign-in, so a finished PDF or a set of images can go out from the same chat where you wrote the caption, after it shows you a preview and waits for your confirmation. Two honest limits worth knowing: document and multi-image posts go through LinkedIn’s newer versioned interface, which may not be switched on for every self-serve app, so try it on your own account rather than assuming; and document posts publish immediately — they cannot be scheduled yet.

A repeatable checklist

• Read the source and write down every claim it makes. Pick one.
• Write ten plain-text lines: claim, stakes, three to five supports, example, action, close.
• Read those lines end to end. Cut anything that does not serve the claim.
• Check each line stands alone for someone who joined mid-swipe.
• Design last, using one template so the set looks like a set.
• Write a new caption. Do not reuse the introduction.
• Park the claims you did not use. They are the next two carousels.

Done this way, one solid article reliably yields three carousels over a couple of months, each a complete argument rather than a third of a summary. That is a better return than squeezing the whole thing into ten crowded slides once, and it is far less work than writing three new pieces from nothing.