5 ChatGPT Prompts Every Marketer Should Know

Most people type one line into ChatGPT, get a bland answer and decide "AI is overrated". The problem is rarely the model. It is the prompt. A marketer who gives ChatGPT the same brief they would give a junior colleague gets usable work back; a marketer who types "write an ad" gets a greeting-card.

Below are five prompt frameworks we use in class and on live client work. Each one follows the same structure: role, context, task, format, constraints. Copy them, fill in the brackets, and keep the ones that work in a notes file so you never start from a blank box again.

1. The ad-copy prompt (with variations built in)

Ad platforms reward testing. Asking for one headline is a waste; ask for a set that you can split-test.

You are a performance marketer writing Meta ads for [product/service] in [city].
Audience: [who they are, one line]. Main pain: [pain]. Offer: [offer].
Write 5 primary texts (max 90 words) and 10 headlines (max 40 characters).
Mix angles: 2 pain-led, 2 benefit-led, 1 social-proof-led.
Avoid: exclamation marks, the word "revolutionary", any claim we cannot prove.
Output as a table: Angle | Headline | Primary text.

Why it works: the character limits stop you from editing later, the angle mix forces variety, and the "avoid" list removes the clichés ChatGPT reaches for first. Always replace the proof points with real ones before the ad goes live.

2. The persona prompt

A persona is only useful if it changes a decision: which platform, which hook, which objection to answer first. Ask for the decisions, not a biography.

Act as a market researcher. Our business: [one line]. Location: [city].
Build 3 buyer personas. For each give: name and age range, what triggers them to look for us,
the 3 objections they raise before buying, where they spend time online (specific apps and content types),
and the single message that would make them click.
Keep each persona under 120 words. Flag anything that is an assumption we should validate with real customers.

The last line matters. ChatGPT will confidently invent details. Asking it to mark assumptions turns the output into a checklist for a 10-minute customer call instead of a fiction you build a campaign on.

3. The 30-day content calendar prompt

You are a social media strategist for [brand] on [Instagram/LinkedIn].
Goal for this month: [leads / followers / launches].
Pillars: [education, behind the scenes, proof, offers].
Create a 30-day calendar as a table: Day | Pillar | Format (reel/carousel/story/post) | Hook (first line) | CTA.
Rules: no more than 1 direct sales post per 4 posts, every reel hook under 8 words, reuse each idea once as a different format.

Then run a second prompt on the output: "Rank these 30 ideas by how likely they are to be saved or shared, and tell me which 5 to film first." Two short prompts beat one giant one.

4. The landing-page rewrite prompt

Paste your existing page copy and ask for a critique before a rewrite. The critique is usually more valuable.

Here is the copy of our landing page: [paste].
First, list the 5 biggest reasons a visitor would leave without enquiring, in order of impact.
Then rewrite only the hero section (headline, sub-headline, button text) in 3 versions:
one clarity-first, one benefit-first, one urgency-first. Keep our tone: plain, confident, no hype.

5. The campaign report prompt

Clients and managers do not want 40 metrics. They want what happened, why, and what you will do next.

Here are this week's numbers for [campaign]: [paste table: spend, impressions, clicks, CTR, CPC, leads, CPL, by ad set].
Write a client update in under 200 words: 3 bullet points on what changed vs last week, 1 clear reason for the biggest change,
and 2 actions for next week with expected effect. Do not restate every number. Plain English, no jargon.

Check the model's arithmetic. ChatGPT reads tables well but still makes percentage errors, so keep the raw sheet open beside it.

Three habits that make every prompt better

• Give it an example. One sample of copy you like will shape the output more than five adjectives.
• Ask for the thinking first. "List the options, then pick one and explain why" produces better choices than "pick one".
• Never publish the first draft. Treat the output as a junior's draft: fix the facts, cut 30 percent, add the proof.

Prompting is a skill you build by doing it on real briefs. In the Digital Marketing Professional Program we use these frameworks on live brand projects every week, so students leave with a prompt library they built themselves, not a PDF of someone else's.