Verified against ChatGPT · 2026-08-09
Break one dense idea into a LinkedIn carousel that survives being read on a phone, one thumb-swipe at a time
Turns a single idea or framework into a slide-by-slide carousel outline — cover hook, one idea per slide, a closing CTA slide — built around how people actually swipe through carousels rather than how a slide deck reads on a laptop.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are outlining a LinkedIn carousel, slide by slide, from one idea I want to teach or make a case for. This is not a slide-deck outline for a meeting — assume every slide is viewed alone, on a phone screen, for about two seconds, by someone deciding whether to swipe to the next one.
CORE IDEA OR FRAMEWORK
A 4-step framework for pricing a freelance retainer so you never underquote a project again.
NUMBER OF SLIDES
8
WHO THIS IS FOR
Freelance designers who currently price by the hour and undercharge.
WHAT PROOF OR EXAMPLE I HAVE
Switched a client from hourly to retainer pricing and increased monthly revenue from that account by 40% without doing more work.
DESIRED ACTION AFTER READING
DM me the word 'retainer' and I'll send the pricing calculator I use.
STRUCTURE RULES
Slide 1 is the cover and carries the entire burden of getting a swipe — it must state a specific, non-obvious claim or promise, not a topic label ("5 tips for X" is a topic label; "the tip that got us fired if we didn't follow it" is a claim). Each interior slide covers exactly one idea, never two ideas competing for the same slide — if the core idea I gave you doesn't cleanly divide into the slide count requested, tell me that instead of padding weak slides with filler. Each slide's text must be readable in under three seconds — that means a short headline plus at most two supporting lines, never a paragraph. Order the slides so each one creates a specific reason to swipe to the next — a question raised, a number teased, an incomplete thought — rather than each slide being a self-contained, disconnected tip that could be read in any order. The second-to-last slide should contain the single most useful, concrete piece of the framework — not save the best part for a slide people may not reach. The final slide is the only slide allowed to ask for an action, and it should ask for one specific thing tied to the end_action given, not a generic "follow for more."
WHAT NOT TO DO
Do not write slide text as full sentences with subordinate clauses — carousels are read, not read aloud; every slide should look like it was cut down, not expanded up. Do not repeat the cover slide's claim verbatim on slide 2 as a way of "recapping" — that wastes the first swipe.
OUTPUT FORMAT
A numbered list, one entry per slide: slide number, the exact text for that slide (headline + supporting lines), and a one-line design note (e.g. "large number, high contrast") only where it meaningfully changes how the slide should look.Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
The instruction to treat each slide as viewed alone for two seconds directly counters the default the model reaches for otherwise: because carousels are visually similar to slide decks in the model's training exposure, an unconstrained request tends to produce slides that read like bullet points from a presentation meant to be narrated, with connecting language ('as we discussed,' 'building on this') that only makes sense if slides are read in sequence with a speaker's context — which is not how a LinkedIn carousel is actually consumed. Requiring the model to flag a mismatch between slide count and how many discrete ideas the core idea actually contains addresses a specific failure mode of forced-length content generation: an LLM asked for exactly N slides will pad thin ideas into N slides rather than pushing back on the count, producing filler slides with no real content, so making the pushback an explicit instruction rather than hoping the model self-corrects meaningfully changes the output. The ordering rule about creating swipe tension is a structural technique specific to how carousels perform on the LinkedIn algorithm, which weights average time-on-post and swipe-through rate; a model with no visibility into that mechanic will default to logically ordering ideas by importance rather than by curiosity gap, so the instruction has to state the actual mechanism (why swipe-inducing order matters) rather than just saying 'make it engaging,' which is unfalsifiable and would be satisfied by any output.
What you get back
Slide 1: "The pricing mistake that cost me $30K before I fixed it." | Slide 2: "Hourly pricing punishes you for getting faster." (large contrast text) | ... | Slide 7: "The exact formula: (hours saved x hourly rate) x 1.5 = retainer floor." | Slide 8: "DM me 'retainer' for the calculator I use with every new client."
Verified against
ChatGPT GPT-5.1 · 2026-08-09
Changelog
- 2026-08-09 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — Google Ads management, built and maintained for you — that's Scult's day job.
EXPLORE GOOGLE ADS MANAGEMENT
