← Blog
·7 min read

The Best ChatGPT Prompts for Copywriters in 2026 (Copy These Today)

Copywriters who treat ChatGPT as a junior writer — giving it context, constraints, and a specific format — produce output that’s worth editing. Those who use it as a magic button produce slop. The difference is entirely in the prompt.

This post covers 10 battle-tested prompt structures for the real work copywriters do daily: headlines, sales pages, email sequences, ads, VSL scripts, and more. Each one is ready to copy, fill in the brackets, and run.

The Anatomy of a Good Copywriting Prompt

Every effective copywriting prompt has four elements. Miss one and the output drifts toward generic. Get all four right and the first draft is usually 70% of the way there.

  1. 1. Role — “You are an expert direct-response copywriter…” This sets the voice, expertise level, and mental model ChatGPT applies to the task.
  2. 2. Context — Product name, target audience, key benefit, tone. The more specific you are here, the less generic the output.
  3. 3. Task — Exactly what to produce: a headline, a subject line, body copy, a script. Don’t leave this implicit.
  4. 4. Constraints — Word count, format, what to avoid. Constraints are what separate a useful draft from a wall of text.

Weak prompt vs. strong prompt

Weak

Write me a headline for my product.

Strong

You are an expert direct-response copywriter. Write 5 headline variations for a productivity app for freelancers. The primary benefit is saving 3 hours a week on admin. Use: curiosity gap, bold claim, and transformation promise formulas. Keep each under 10 words.

The strong prompt gets 5 specific, structurally different headlines. The weak prompt gets one forgettable line. Same tool, completely different output — because of the brief.

10 Copy-Ready Prompts for Every Format

1. The Direct-Response Headline Generator

Use case: Product/landing page headlines

The Prompt

You are an expert direct-response copywriter. I'm writing a headline for [PRODUCT]. Target audience: [WHO THEY ARE]. Primary benefit: [MAIN BENEFIT]. Write 10 headline variations using these formulas: curiosity gap, numbered list, transformation promise, bold claim, and question. Keep each under 12 words.

Why it works: Generating across five formulas simultaneously forces variety. Instead of polishing one angle, you get 10 structurally different options — then pick the strongest or A/B test two.

2. The PAS Framework Body Copy Writer

Use case: Sales pages, product descriptions

The Prompt

You are a conversion copywriter. Write a 3-paragraph PAS (Problem → Agitate → Solution) block for [PRODUCT]. The reader is [AUDIENCE DESCRIPTION]. Problem: [PROBLEM]. Agitate by making the pain vivid and specific. Solution: position [PRODUCT] as the relief. Tone: [TONE]. Keep each paragraph under 80 words.

Why it works: The Agitate instruction is the key lever. Without it, ChatGPT defaults to a light acknowledgment of the problem and jumps straight to solution. Explicitly telling it to make the pain vivid produces copy that makes the reader feel seen — and that's when they start paying attention.

3. The Email Subject Line Machine

Use case: Email marketing

The Prompt

You are an email copywriter known for high open rates. Write 15 subject line variations for an email about [TOPIC/OFFER]. Audience: [WHO]. Include: 5 curiosity-based, 5 benefit-driven, 5 pattern-interrupt. Flag which 3 you'd A/B test first and why. No emojis unless specifically requested.

Why it works: Asking for 15 across three distinct categories prevents ChatGPT from defaulting to one register. The “flag your top 3” instruction is useful too — it forces the model to evaluate its own output, which often surfaces the best options faster.

4. The Facebook/Instagram Ad Copy Generator

Use case: Paid social ads

The Prompt

You are a performance marketing copywriter. Write 3 variations of Facebook ad copy for [PRODUCT]. Audience: [WHO]. Goal: [click to product page / sign up / etc.]. Format each as: Hook (first 125 chars, stops the scroll) → Body (builds desire, max 200 chars) → CTA (one sentence). Tone: [TONE].

Why it works: The 125-character hook constraint mirrors how Facebook actually truncates copy before the “See More” break. Writing to that constraint forces a scroll-stopping opener, not a slow wind-up.

5. The Long-Form Sales Page Opener

Use case: Sales page above-the-fold section

The Prompt

You are a long-form direct-response copywriter in the style of Gary Halbert. Write the opening section (hook + subhead + first 3 paragraphs) of a sales page for [PRODUCT]. Target reader: [WHO]. They are currently experiencing [PAIN POINT]. Open with a story or bold claim that makes them feel seen. No corporate language. Conversational, punchy.

Why it works: Style attribution (“in the style of Gary Halbert”) meaningfully shifts the register away from AI-sounding corporate copy. The explicit instruction to make the reader feel seen before selling anything is the single biggest conversion variable on long-form pages.

6. The VSL Script Intro

Use case: Video sales letters, YouTube ads

The Prompt

You are a VSL scriptwriter. Write the first 90 seconds (approx. 225 words) of a video sales letter script for [PRODUCT]. Audience: [WHO]. Hook them in the first 10 seconds with [PAIN POINT or BOLD PROMISE]. Then build curiosity about [TRANSFORMATION]. Tone: [TONE]. Format as a script with [PAUSE] and [EMPHASIS] notes.

Why it works: The [PAUSE] and [EMPHASIS] formatting instruction makes the script immediately usable — you’re not translating prose into something speakable. The 10-second hook constraint is non-negotiable: YouTube viewers skip at 5 seconds, and you need to earn every second after that.

7. The Product Description Writer

Use case: eCommerce product pages

The Prompt

You are a product copywriter. Write a product description for [PRODUCT NAME]. It [WHAT IT DOES]. Target buyer: [WHO]. Lead with the top emotional benefit, then list 5 feature-benefit bullets (format: Feature — why it matters to the buyer). Close with a single sentence CTA. Max 200 words total. Tone: [TONE].

Why it works: The feature-benefit bullet format (“Feature — why it matters”) prevents the spec-sheet mistake most product descriptions make. Buyers don’t buy features; they buy what those features do for them. The format enforces that distinction automatically.

8. The Nurture Email Writer

Use case: Email sequences

The Prompt

You are an email copywriter. Write email #[NUMBER] of a [NUMBER]-email nurture sequence for [PRODUCT/BRAND]. Context: subscriber [signed up for / downloaded / purchased]. Goal of this email: [educate / build trust / introduce offer]. Tone: [TONE]. Keep it under 250 words. End with one clear next step, not a hard sell.

Why it works: Specifying the email’s position in the sequence matters more than most people realize. Email #2 of a 5-email sequence has different stakes than email #4. Giving ChatGPT that context produces copy calibrated to the right moment in the buyer journey, not just generic nurture filler.

9. The Objection Crusher

Use case: Sales pages, FAQ sections, ads

The Prompt

You are a persuasion copywriter. The top 3 objections prospects raise about [PRODUCT] are: [OBJECTION 1], [OBJECTION 2], [OBJECTION 3]. For each objection, write a 3–4 sentence response that acknowledges the concern, reframes it, and pivots to a benefit. Tone: empathetic but confident, not defensive.

Why it works: Objection handling is the most skipped section of sales copy. Most writers address the product; this prompt addresses the doubt. “Empathetic but confident, not defensive” is a critical tone instruction — defensiveness signals that the objection is valid. Confidence without dismissiveness is what converts.

10. The Testimonial Polisher

Use case: Turning rough customer quotes into polished social proof

The Prompt

You are a copywriter specializing in social proof. I have a rough customer testimonial: '[PASTE RAW QUOTE]'. Polish it into a punchy, specific, believable testimonial of 2–3 sentences. Keep the customer's voice, make the result specific (add numbers if implied), and cut vague language. Do not invent anything — only sharpen what's there.

Why it works: Raw testimonials are usually too long, too vague, or buried in qualifiers. This prompt tightens without fabricating. The “do not invent anything” constraint is explicit for good reason: believable social proof is specific and honest. Any inflation is detectable and kills trust.

How to Use These Prompts Effectively

The prompts above are starting structures, not magic incantations. How you use them determines whether the output is a useful first draft or more work to fix than starting from scratch.

  • Fill in the brackets before running. Context is the biggest lever you have. A vague fill-in produces a vague output. The more specific you are about the audience, the product, and the tone, the better the first draft.
  • Treat first output as a draft. Edit it like you’d edit a junior writer: keep the structure, sharpen the language, add any specific details ChatGPT couldn’t know about. You’re the editor, not the typist.
  • Iterate. If the first output misses the mark, don’t restart — refine. Try: “Make the hook more aggressive,” “Rewrite in a more casual tone,” or “Give me 3 more variations of the CTA.” The conversation is the tool.
  • Stack prompts. Use the headline generator first, then feed the winning headline into the sales page opener. Use the PAS block as the body of the ad copy. Each output becomes the brief for the next prompt.

Want 100+ more prompts like these?

The Ultimate ChatGPT Prompt Pack covers headlines, sales pages, email sequences, ads, social, and more — tested across every major copy format. Instant download, yours to keep.

Get the Prompt Pack — $19 →

ChatGPT doesn’t replace copywriters. It replaces the blank page. The skill is still the same — understanding the reader, the product, and the moment. The prompts above just compress the time between brief and first draft. Everything after that is craft.

Ready to move faster with AI?

Flux sells battle-tested prompt packs and no-code automation templates.

Browse the store →