ChatGPT Prompts for Marketing: The Library

There are hundreds of “ChatGPT prompts for marketing” lists. Most of them are copy-paste dumps: fifty prompts, zero explanation, and no way to tell why one works and another produces generic filler. Copy a prompt you don’t understand and you can’t fix it when it breaks - and it will break, the moment your product, audience or brand voice differs from whatever the list-writer imagined.

I built this library the opposite way. Fewer prompts, each one explained: what it asks for, why I structured it that way, and how to adapt it before you run it.

What makes a marketing prompt work

Every prompt in this library uses the same four-part anatomy. Steal it - I lean on it for almost every prompt I write from scratch:

1. Role - who the model should act as (“You are a lifecycle marketer at a B2B SaaS company”). This sets vocabulary, priorities and taste.
2. Context - the facts it can’t guess: your product, audience, offer, tone, and one or two examples of past work you liked.
3. Constraints - the rules of the format: character limits, banned phrases, required structure, how many variants you want.
4. Output format - exactly what the response should look like, so you can use it without reformatting.

Most bad AI output is a missing-context or missing-constraints problem, not a model problem. When a prompt disappoints, fix parts 2 and 3 before trying a different tool.

How to adapt any prompt here

Before running a prompt from my library, replace the bracketed placeholders with your own product and audience, then add one example of a past asset you consider good. That single example does more for output quality than any clever wording - it shows the model your bar instead of describing it. Why examples work that well, how many to use and the biases to watch for is covered in the few-shot prompting guide.

When the output is good enough

AI output is a draft, not a deliverable. My working rule: ship it when you’d have accepted the same draft from a competent freelancer - and edit it exactly as hard as you’d edit that freelancer. If you’re accepting output you wouldn’t accept from a person, the prompt isn’t saving you time. It’s lowering your bar.

The prompt guides

ChatGPT prompts for blog posts - ten outline-and-draft templates grouped by job (ideation, outlining, drafting, hooks, refreshing old posts), each with its reasoning and adaptation notes.

ChatGPT prompts for email subject lines - prompts that generate A/B-ready subject-line pairs with mobile truncation limits built into the prompt itself, organized by campaign type.

ChatGPT prompts for email newsletters - seven prompts on a story, tip and CTA rhythm, with an anti-invention rule that keeps the model shaping your material instead of fabricating it.

ChatGPT prompts for product descriptions - a fact-sheet system that scales descriptions across SKUs and channels without the invented claims that drive returns.

ChatGPT ad copy prompts for Google, Meta and LinkedIn - one template per platform, because a prompt that ignores Google’s 30-character headlines or LinkedIn’s truncation writes ads you can’t ship. Includes a compliance guardrail block.

ChatGPT prompts for social media captions by platform - one prompt per platform with the fold and the hook built in, plus the repurposing and results-review prompts that turn one post into a week of captions.

ChatGPT prompts for high-retention YouTube scripts - seven prompts built on the retention report YouTube actually gives you: the hook written against the thumbnail, a payoff schedule instead of a bait-and-delay, the cut-only dip audit, and the post-mortem that reads the graph against the script.

ChatGPT prompts for landing page copy - seven section-by-section prompts built on message match: the hero, benefits, objections, proof and the copy around the form, each running on real inputs with a proof fence the model may not cross.

ChatGPT prompts for buyer personas that feel real - seven prompts that synthesise personas from real customer evidence instead of letting the model invent Marketing Mary, with jobs-to-be-done and an anti-fabrication rule built in.

ChatGPT prompts for competitor analysis - seven prompts that turn reviews, pricing pages, ad libraries, job posts and your own win/loss notes into evidence-cited intelligence and a battlecard, without letting the model invent your rival.

ChatGPT prompts for case studies that convince - six prompts from interview transcript to approved story: evidence extraction with the friction kept in, the angle decision, results framed without invented arithmetic, and quotes edited by cuts only.

Few-shot prompting for marketers - the technique under the whole library: why 3-5 pasted examples beat paragraphs of instruction for voice and format, where to source them, and the ordering biases the research warns about.

Brand voice AI prompts: train ChatGPT on your guidelines - the setup every other prompt here assumes: turn adjectives into checkable rules, add three examples, and run a voice-check prompt on every draft.

System prompts for ChatGPT marketing consistency - the standing layer under everything above: what belongs in the system prompt versus the task prompt, the four places to install it in ChatGPT, and the template with product facts and claims rules built in.

Claude Projects for marketing: repeatable AI workflows - the same standing-context idea on Claude’s side: instructions separated from a file-based knowledge base, the four Projects worth building, and team sharing as enforced consistency.

Build a custom GPT for email marketing - the standing setup as a shareable tool: the full instructions template, the three knowledge files that earn their place, and the historical-brief test that says whether the build works.

ChatGPT fact-checking prompts: verify every claim - six prompts that make the model list, sort and stress-test every claim in a draft - extraction, source audits, quote checks, an adversarial pass - while the verdict stays human.

ChatGPT prompts for SEO content briefs - eight prompts that build a brief piece by piece: intent read from the real SERP, gaps, thesis, structure, internal links, and a QA pass that says whether the brief is too thin to bother drafting from.

Build an AI content brief workflow that scales - the workflow layer above individual prompts: how briefs, not drafts, decide whether AI content is any good.

Prompt chaining in ChatGPT to bulk-generate content - the pattern for volume: angles, outline, draft and polish as separate prompts with a review gate between each, and the batch discipline that keeps piece twenty as good as piece one.

Multi-turn prompting in ChatGPT for iterative work - the unscripted counterpart to chaining: the five rules that make a refinement loop converge, two worked loops, and the signs a conversation is spent.

Content calendar workflow: plan a quarter with ChatGPT - themes chosen by you and challenged by the model, topic benches generated against what you already published, scheduling at honest capacity, and the weekly fifteen-minute review that keeps the calendar alive.

Content repurposing workflow: automate it with ChatGPT - one source post into a week of derivatives: the extraction pass that separates real repurposing from lazy translation, the one-claim-per-post rule, and the two steps that never get automated.

From meeting notes to campaign plan - the chain for messy inputs: extract what was actually said, find the gaps, confirm the record with the room, and only then let the model build the brief and the plan.

For the bigger picture of where prompts fit in an AI marketing practice, start with the complete guide to AI marketing.