I've posted on LinkedIn for three years and for most of that time my carousels averaged 3,000 to 5,000 impressions. Respectable for my audience size, forgettable in every other way — the content equivalent of a polite golf clap from across a quiet room. Then in March I posted one carousel that hit 127,000 impressions and pulled in 620 new followers in a single week. People immediately asked what the secret topic was. There wasn't one. The topic — mistakes new managers make in their first 90 days — was not original. The framework inside it wasn't either. The only thing that changed was how I scripted the slides, and it's so simple it almost annoys me to admit it.
The structural disease my old carousels had
When I went back and looked at my old scripts with fresh eyes, the problem was embarrassingly obvious. I crammed. My hook slide tried to explain the entire concept on slide one, which meant there was no reason on earth to swipe to slide two — I'd already given away the plot in the trailer. My teaching slides were mini-essays, three dense paragraphs each, the kind of thing you screenshot to 'read later' and then never, ever do. Swiping through one of my old carousels felt like reading a PDF a stranger emailed you, not following a story that physically pulls you to the next frame.
A carousel is not an article cut into ten pieces. I had been making articles cut into ten pieces for three years and quietly wondering why they refused to travel.
What the prompt forced me to do differently
The Swipe-Worthy Carousel — A 10-Slide LinkedIn Script From Any Idea prompt enforces the exact opposite discipline from my worst habits: one idea per slide, body copy capped at 50 words, and a specific visual direction for every single frame. When I ran '7 mistakes new managers make in their first 90 days' through it with Claude Opus 4.8, each slide came back feeling like a breath instead of a brick. There was white space. There was a single clean thought per frame. There was a deliberate reason, built into every slide, to swipe to the next one rather than bounce.
Why Claude Opus 4.8 specifically
Carousel scripting is a structure-and-restraint problem far more than a wordsmithing one. You need a model that will actually hold the line on 'slide 1 is a hook with zero preamble, slides 2 through 8 are one numbered idea each, slide 9 is a screenshot-able summary, slide 10 is a low-friction CTA' without quietly drifting back into essay mode by slide four. Claude Opus 4.8 respects that scaffolding with genuine discipline, and the light XML tags in the prompt give it clean, labeled lanes to stay inside. The output came back orderly — no slide trying to secretly do two jobs, no slide nine sneaking in a brand-new idea that should have been its own frame.
The specific choices that moved the numbers
- Slide 1 hook: 'Most new managers lose their best people in the first 90 days. Here's why.' A discomfort hook the algorithm visibly rewards with reach.
- Each teaching slide was exactly one mistake, one sentence of explanation, and one visual — a bold number on a colored block. Nothing to read, everything to absorb in under five seconds.
- The summary slide was built from the ground up to be screenshotted: three clean bullets, generous spacing, zero clutter. People save and reshare what they can screenshot in one tap.
- The companion caption teased the payoff without spoiling it, so the comments filled with people who'd actually swiped all the way through rather than reactors who never opened it.
The companion caption is the part everyone skips
Here's the detail almost nobody pays attention to, and the one I now think mattered nearly as much as the slides themselves: the companion caption the prompt writes to post alongside the carousel. On LinkedIn, the caption is what shows in the feed before anyone opens the document, so it's the actual hook for the whole thing — the carousel could be perfect and still die if the caption gives the game away. The prompt is explicitly told to tease without repeating, and that single instruction stopped me from doing what I'd always done, which was paste the first slide's text into the caption and wonder why nobody swiped. A good caption sells the click into the carousel; a bad one makes the carousel redundant. Mine now ends on a small open loop, and the open rate on the document itself jumped noticeably once I stopped spoiling it.
The carousel took about an hour to design in Canva once the script existed. The script itself took ten minutes to generate and fifteen to edit. My previous all-time best for any carousel was 22,000 impressions; this one did nearly six times that. Structure was the only variable I changed, and the prompt enforced the structure I'd been resisting for three years precisely because it felt too simple to be the answer. It is simple. That's the entire, slightly humbling point. Grab the Swipe-Worthy Carousel — A 10-Slide LinkedIn Script From Any Idea prompt on Prompt Dock, feed it a tip you've already posted as a boring text update that went nowhere, and watch it become something people swipe all the way to slide 10. When I need fresh angles for the next one, I run the topic through my Ten Hooks, One Topic prompt before I even open the carousel scripter.