Formats
LinkedIn carousel posts that get engagement: structure, length and hooks
· 8 min read · Dave Banerjee
Carousels earn reach through dwell time. Every swipe is a few more seconds on your post, and a document people finish is one of the strongest signals you can send. But a bad carousel is worse than a text post, because the cover slide has to do all the work of a hook with none of the words.
The cover slide is the whole hook
Six to nine words, high contrast, one promise. "5 emails that got replies from cold prospects" works. "Thoughts on outbound" does not. If your cover needs a subtitle to make sense, it isn't a cover yet.
Length and structure
- 8 to 12 slides. Under 6 feels thin; over 14 and the completion rate falls off a cliff.
- One idea per slide, one sentence of support. Slides are not paragraphs.
- Number the slides so readers know how much is left — it measurably increases finishes.
- Square (1080×1080) or 4:5. Landscape wastes the mobile screen.
The shape that keeps people swiping
- Cover: the promise.
- Slide 2: the cost of not knowing this — why keep swiping.
- Slides 3–10: one point each, concrete, with a real example or number.
- Second-to-last: the summary somebody could screenshot.
- Last: the ask — a question, not a pitch.
Design constraints that actually matter
Big type (28pt minimum on a 1080px slide), dark text on light or a single strong brand colour, and generous margins so nothing is clipped in the feed preview. No stock photography behind text. Consistency across your carousels matters more than any individual layout — people should recognise your slides before they read the name.
The caption still does half the job
Write the caption like a text post: two lines that make the swipe feel worth it, then the question you want answered in the comments. A carousel with no caption gets scrolled past by everyone who does not tap images.
One last thing that separates carousels that work from carousels that get made: reuse the ones that land. Take your best-performing text post, turn each of its points into a slide, and publish it again three weeks later. The audience overlap is smaller than you think, and the format does the rest.