I built a viral carousel using a Claude Skill. Here is the full system, free.

The hook on the first was “how I booked 30 appointments in a month using this strategy,” and the last slide said comment “outbound” to get the full breakdown. That failure taught me more than the win did, and it is the reason this piece is worth reading. Because everyone tells you the hook is everything, and it is not. The whole system, including the free Claude Skill I use to build it, is below.

Why do some carousels get saved and shared while others die?

Not the hook. The hook only earns the swipe. What gets a carousel saved and shared is what happens after it.

I learned this the hard way. My first carousel worked, and like everyone else I assumed the hook was the reason. So on the next one I made sure the hook was just as strong. Same promise structure, same clarity, same scroll-stopping first slide. It died. Low saves, low shares, weak reach. I could not work out why, because the one thing everybody obsesses over was as good as it had ever been.

Here is what was actually different. The first carousel had specific, valuable middle slides: the real tools, how we used them, the actual context, one clear point per slide. The last slide was one clean line, comment “outbound” to get this in your DM, and nothing else. The second carousel had a great hook and then fell apart. Each slide had too much crammed in, the points were vague instead of specific, and the final slide was a mess of CTA plus logo plus follow plus social handles plus contact info. People swiped in on the hook, hit a wall of unfocused text, and dropped.

The hook gets them in. The middle keeps them moving. The clean ending gets them to act.

Why the middle slides decide your reach

Because completion is the signal, and the middle is what drives completion.

LinkedIn reads every swipe as sustained attention, and the further people get through your carousel, the harder the platform pushes it. When a high share of viewers reach the final slide, the post gets driven into the feeds of people who do not follow you. That reach does not come from the hook. It comes from every middle slide being good enough to earn the next swipe. So a vague, overloaded middle does not just bore people, it actively kills your distribution, because it stops the completion the algorithm is rewarding.

This is also why saves matter so much right now. Saves are the highest-value signal in the current LinkedIn algorithm, because a save tells the platform this is reference-worthy material, and it pushes the post wider. People save carousels that are specific and actionable, the ones that feel like a free resource they will want again. They do not save vague. My first carousel was a save-worthy breakdown someone could actually use. My second was noise with a good title.

The system I use to build them

A repeatable structure, plus a Claude Skill that produces the slides, with me still making the calls that matter.

The backbone is a simple five-to-seven-slide framework, and it never changes. Slide one is the hook. Slide two sets up why it matters. The middle slides carry one specific point each, never two, because the moment you cram two ideas onto a slide you lose the clarity that earns the swipe and the save. The final slide is one clean call to action and nothing else.

The Claude Skill I built runs that framework. I give it the topic and the angle, and it generates the slides to the structure, one idea per slide, within tight readability limits so no slide ever becomes a wall of text. It handles the consistency automatically, same layout, same fonts, same spacing across every slide, which is what makes a carousel look like one considered piece rather than seven random images. What the skill does not do, and what I still do myself, is decide the hook and make the final calls on design emphasis. The system produces a clean, correct, consistent carousel every time. The judgment about which outcome to lead with, and how hard to push the hook, stays human. That is the honest division of labour, and it is why the system is reliable without being on autopilot.

One technical note, because it trips people up. LinkedIn removed native carousels at the end of 2023, so a carousel now is a PDF you upload as a document post. That is the format, and it is currently the single highest-engagement format on the platform, sitting well above text and video, while fewer than two percent of posts actually use it. It is a wide-open lane.

What most people get wrong about carousel design

They lead with the design. The design-first crowd obsesses over colours, fonts, and slide count, and they have it backwards.

I used to build beautifully designed carousels. For one travel client I made a genuinely polished one, and it did not perform. When I stopped leading with design and started leading with the outcome, the hook, the specific content, and a strong clean CTA, that is when it worked. Design is not irrelevant, but its job is narrow: make the hook and the outcome impossible to miss, and keep everything else clean and uncluttered so nothing competes for attention. A gorgeous carousel with a weak hook and a vague middle loses to a plain one with a sharp hook and specific, useful slides every time.

So the design rules are simple. Big, bold, readable font on the hook and the outcome. One clear point per slide. Consistent look across all slides so it reads as one piece. And a final slide with a single ask, not a billboard of every way to contact you. Clean beats clever, because clean keeps the focus on the thing that actually drives the save: the value.

Build your own, and here is the whole thing free

Here is the full system in order, so you can run it yourself.

1. 01

   Pick one specific outcome your audience wants and make it the hook. Not a topic, an outcome. “How I booked 30 appointments in a month” beats “tips for outbound” because the reader instantly wants that result for themselves.

2. 02

   Write five to seven slides on the framework. Hook, then setup, then one specific point per slide through the middle, then one clean CTA. If a slide has two ideas, split it. If a slide has vague advice, make it concrete or cut it.

3. 03

   Keep every slide under about forty words. Keep the design consistent and uncluttered, so people can move through it fast and nothing fights the message.

4. 04

   Make the last slide a single action. Comment one word to get something in their DM works because it is easy, specific, and it opens a conversation. Do not stack five asks on it.

5. 05

   Post it as a PDF document upload. Write a caption that ends with a question, and then reply to every comment in the first hour, because early engagement is what tells the algorithm to push it wider.

Do one thing this week if you have a carousel that flopped. Go back and look at it, and ignore the hook. Read the middle slides as a stranger and ask whether each one is specific and worth the swipe, then look at the last slide and count how many things you asked for. That is almost always where it broke, and it is the cheapest fix you will find.