ChatGPT Prompts for Copywriters

The best ChatGPT prompts for copywriters give the model the same information you would give a skilled copywriter: audience, offer, objective, proof, voice, and constraints. With that context, ChatGPT can help you research, generate ideas, draft variations, and edit copy. But its output still needs a copywriter’s judgment, fact-checking, and understanding of the customer.

This guide is a practical prompt library for real copywriting tasks—not just a list of generic requests. You’ll find prompts for research, headlines, landing pages, sales pages, product descriptions, ads, email, calls to action, brand voice, editing, and A/B test ideas. Each prompt uses editable placeholders so you can adapt it to your own brief.

Replace those placeholders with verified information wherever possible. If you don’t have customer research, product proof, or approved brand examples, ask ChatGPT to identify what’s missing instead of filling in the gaps. AI-generated copy can sound confident even when it makes assumptions, so treat every output as a draft to review.

What makes a good ChatGPT prompt for copywriting?#

A useful copywriting prompt does more than name the deliverable. “Write a sales email” leaves out who the email is for, why they should care, what they should do next, and which claims the writer is allowed to make. The result may be fluent, but it’s likely to rely on generic messaging or invented details.

A stronger prompt gives ChatGPT a working brief. OpenAI’s prompting guidance recommends being clear and specific, providing relevant context, describing the desired format or tone, and refining prompts iteratively rather than expecting a perfect first response. For copywriting, use this framework:

• Context: What is the product, offer, market, funnel stage, and campaign goal?
• Audience: Who is the copy for? Include relevant pains, desires, objections, and awareness level.
• Evidence: Supply verified features, benefits, differentiators, customer language, and proof.
• Task: Specify the exact deliverable—such as five subject lines, a hero section, or three ad variations.
• Voice: Describe the desired tone, vocabulary, sentence style, and phrases to avoid. Approved samples are even more useful.
• Constraints: Set length, format, channel requirements, required wording, and prohibited claims.
• Evaluation: Ask for alternatives, explain what information is missing, or check the draft against your brief.

Use a clear instruction at the beginning, then separate your instructions from source material with labels such as BRIEF, CUSTOMER RESEARCH, and APPROVED CLAIMS. If a prompt includes sensitive customer information, follow your organization’s privacy and data-handling rules before sharing it.

A general prompt structure looks like this:

You are helping me create [asset] for [audience]. The offer is [offer]. The goal is [goal]. Use the customer evidence and approved proof below. Write in [tone] and follow these constraints: [constraints]. Do not invent product capabilities, statistics, testimonials, customer results, pricing, guarantees, certifications, or scarcity. If critical information is missing, ask questions before drafting.

Then use follow-up prompts to improve the result. You might ask for a clearer version, a different angle, or a more specific benefit—but don’t ask the model to make a claim “stronger” unless you’ve supplied evidence that supports it.

ChatGPT prompts for copywriting research#

ChatGPT can help you organize research you provide: interview transcripts, survey responses, reviews, support tickets, sales notes, or product documentation. It cannot automatically know what your customers said in a private interview or what your analytics show. Provide the relevant source material, and ask the model to separate direct evidence from its interpretation.

1. Extract voice-of-customer language

Analyze the customer research below. Extract recurring pains, desired outcomes, objections, buying triggers, exact customer phrases, and evidence or proof points. Separate direct quotes from your interpretation. Cite the source passage for each finding where possible. Do not invent customer insights or treat one comment as a universal pattern.

Research: [paste approved source material]

How to use it: Keep exact quotes intact and record where they came from. The model’s categorization is a starting point; verify patterns against the original material.

2. Find objections and unanswered questions

Review this research for objections or unanswered questions that could affect a purchase of [offer]. Group them by theme and include the exact supporting quote or source for each. Label any possible implication as an interpretation, not a customer statement. Do not create objections that are not present in the material.

3. Turn research into a message map

Using only the research and offer details below, create a message map with: audience problem, desired outcome, current alternative, key differentiator, supporting evidence, likely objection, and a response that can be supported. Mark missing evidence clearly.

Offer details: [details] Customer research: [research]

4. Review competitor copy provided by the user

Compare the competitor copy below with our approved positioning. Summarize themes, audience promises, and claims each competitor emphasizes. Do not infer market share, performance, or customer sentiment. Identify possible points of distinction for our brand, but label them as hypotheses that need validation.

Our positioning: [positioning] Competitor copy: [paste publicly available copy and source]

5. Translate product facts into customer questions

Based on these verified product details, list the questions a prospective [audience] might ask before choosing [product]. For each question, identify which supplied product fact could help answer it. Do not make up capabilities or assume customer priorities that aren’t supported by the research.

ChatGPT prompts for headlines and hooks#

Headlines and hooks are easy to generate in volume and harder to judge well. Give ChatGPT a clear audience, desired outcome, differentiator, and proof. Ask for distinct approaches rather than 20 near-identical variations.

6. Generate headlines by angle

Create 15 headline options for [offer] aimed at [audience]. The primary desired outcome is [outcome]. The main differentiator is [verified proof or difference]. Produce five benefit-led, five problem-aware, and five curiosity-led options. Avoid hype, vague superlatives, and unsupported claims. Label each option by angle.

7. Create landing-page headline options

Write 10 landing-page headline and subheadline pairs for [offer]. The audience is [audience], and the page goal is [goal]. Use this approved promise: [promise]. Keep each pair specific and easy to understand. Do not add guarantees or results beyond the evidence provided.

8. Write ad hooks for different awareness levels

Create 12 short hooks for [offer]: four for people who are problem-aware, four for people comparing solutions, and four for people already familiar with our product. Use only these facts and proof points: [facts]. Keep the audience and awareness label with each hook.

9. Generate email subject lines

Write 15 subject lines for an email about [topic or offer] to [audience]. Group them into: direct, benefit-led, question-based, and curiosity-led. Avoid false urgency, misleading “reply” or “forward” language, and claims not supported by the email content. Keep each under [length] characters.

10. Improve a headline without changing its claim

Evaluate this headline for clarity, relevance, specificity, and fit with the audience. Suggest five alternatives that preserve its factual meaning and do not add stronger claims. Briefly explain the most important difference between the options.

Headline: [headline] Audience and offer: [details]

ChatGPT prompts for landing pages#

A landing page is made of connected sections, not one block of copy. Drafting section by section makes it easier to check message hierarchy, evidence, and consistency. A practical workflow is: build a brief, rank the messages, outline sections, draft each section, address objections, refine the CTA, then review the full page.

11. Build a landing-page message hierarchy

Using the copy brief below, create a landing-page message hierarchy. Rank the primary promise, supporting benefits, proof, objections, and CTA. Explain what each section of the page must accomplish before drafting any copy. Mark missing information as a question rather than filling it in.

Brief: [paste brief]

12. Draft the hero section

Draft three hero-section options for [offer] aimed at [audience]. Each should include a headline, subheadline, and CTA. Base the promise only on [approved promise and proof]. Make the main benefit clear without exaggeration. After the options, list any claim that needs verification.

13. Explain the problem without manufacturing fear

Write a concise problem section for a landing page about [offer]. Reflect these customer challenges: [research-backed challenges]. Keep the tone [tone]. Avoid implying that every reader has the same problem or escalating the stakes beyond the evidence.

14. Present the solution and benefits

Draft a solution section explaining how [product or service] helps [audience] address [problem]. Use these verified features: [features]. Connect each feature to a plausible customer benefit, and show the feature that supports it. Do not imply capabilities or outcomes that aren’t supplied.

15. Place proof where it supports the message

Review the landing-page outline below and recommend where each supplied proof point would be most relevant. Explain what claim the proof can support and what it cannot prove. Do not create testimonials, statistics, case-study results, or certifications.

Outline: [outline] Verified proof: [proof]

16. Address page objections

Using the audience research below, draft a short objection-handling section for [offer]. Address only objections supported by the research and respond using approved product information. If the answer is not available, write “[answer needed]” instead of guessing.

Research: [research] Approved information: [information]

17. Draft a landing-page FAQ

Create up to eight landing-page FAQ questions for [offer] based on the supplied customer questions and product information. Answer only what the information supports. Flag questions that need an internal answer. Do not invent terms, refund policies, compatibility, pricing, or guarantees.

18. Review the finished page for consistency

Review the landing page below against the brief. Check whether the promise, audience, terminology, proof, tone, and CTA remain consistent from top to bottom. Identify unsupported claims, repeated sections, missing answers, and any mismatch between a headline and its evidence. Do not rewrite yet.

Brief: [brief] Page: [draft]

ChatGPT prompts for sales pages#

Sales pages often make more claims and handle more objections than short-form copy. Ask ChatGPT to organize the offer and the argument, but require it to use only supplied evidence. In particular, review any urgency, scarcity, comparison, guarantee, or results language before publication.

19. Frame the offer

Summarize the offer [offer] for [audience] in three ways: a plain-language description, the main customer benefit, and the reason it may be a fit for the audience. Use only these offer details: [details]. Clearly label any benefit that is an inference rather than a stated product fact.

20. Outline a long-form sales page

Create a long-form sales-page outline for [offer] aimed at [audience]. Include the purpose of each section, the claim it needs to make, and the evidence needed to support it. Include sections for the problem, solution, benefits, proof, objections, offer details, and CTA. Do not draft persuasive claims without the supporting information.

21. Map objections to evidence

Organize these objections by theme and match each one to an answer supported by the supplied product information. If there is no adequate answer, mark it “needs product or legal review.” Do not invent a guarantee, policy, result, or explanation.

Objections: [research] Approved information: [details]

22. Review urgency and scarcity

Audit this sales-page draft for urgency or scarcity language. List every statement that implies limited availability, a deadline, a price change, or a special condition. For each, identify the evidence required to substantiate it. Do not make unsupported urgency sound more convincing.

Draft: [copy] Confirmed offer terms: [terms]

23. Plan CTA sequencing

Review this sales-page outline and suggest where calls to action should appear based on the reader’s likely questions and information needs. Explain the purpose of each CTA and whether it should ask for a purchase, trial, demo, consultation, or another next step. Do not assume an action is available unless listed here: [available actions].

ChatGPT prompts for product descriptions#

Product copy must explain value without changing the facts. This is especially important for ecommerce listings, SaaS feature pages, service descriptions, and marketplace listings, where a small invented detail can mislead a buyer.

24. Translate features into benefits

Turn these verified product features into customer-centered benefits for [audience]. Keep technical facts accurate. For each benefit, show the source feature and explain the connection. Do not add capabilities, outcomes, or compatibility that are not in the product information.

Features: [verified features]

25. Write an ecommerce product description

Draft an ecommerce product description for [product] aimed at [audience]. Include a clear opening, the most relevant verified features, practical benefits, and a concise close. Use this product information only: [details]. Do not invent materials, dimensions, certifications, use cases, shipping terms, or performance claims.

26. Write SaaS feature-page copy

Draft copy for the [feature] section of a SaaS product page. Explain what the feature does, who may find it useful, and how it works using these verified details: [details]. Distinguish what the product does today from planned or unavailable functionality.

27. Describe a professional service

Write a service description for [service] aimed at [audience]. Explain the scope, process, deliverables, and next step using the information provided. Do not promise specific outcomes, timelines, or results unless they are explicitly approved here: [details].

28. Adapt product copy to a marketplace format

Rewrite this product description for [marketplace or listing format]. Preserve all factual details, follow the supplied character and formatting constraints, and remove language that is not relevant to this channel. Flag any claim whose accuracy or marketplace eligibility needs human review.

Description: [copy] Requirements: [requirements]

ChatGPT prompts for ad copy#

Good ad development starts with a clear offer and customer insight. Use ChatGPT to explore angles, write variants, and create testable concepts; then review claims against the relevant platform rules and your own evidence. Google Ads, for example, prohibits false, misleading, or unrealistic claims. Platform policies can change, so verify current requirements before launch.

29. Generate ad angles

Generate 10 ad angles for [offer] and [audience]. For each, provide the core idea, customer tension, promise, proof needed, and a sample headline. Do not invent results, guarantees, testimonials, or product features. Mark angles that depend on evidence we have not supplied.

30. Write paid social variations

Create five paid social ad variations for [offer] aimed at [audience]. For each, provide a hook, primary text, headline, and CTA. Use this verified offer and proof: [details]. Vary the message angle, not just the wording. Avoid personal-attribute assumptions and unsupported claims.

31. Draft search ad concepts

Create [number] search-ad concepts for [offer] and these search themes: [themes]. Include a headline idea, description idea, and landing-page message match note for each. Respect these character limits: [limits]. Do not imply availability, price, results, or features not confirmed here.

32. Develop display-ad concepts

Suggest six display-ad concepts for [offer]. For each, provide the visual idea, short headline, supporting line, and CTA. Keep the message understandable at a glance and consistent with the landing page. Use only these approved claims: [claims].

33. Review ad claims before testing

Review this ad draft for factual or potentially misleading claims. Quote each claim that needs checking, state what evidence or clarification is needed, and suggest a neutral alternative only if the supplied facts support it. Do not certify the ad as compliant.

Draft: [ad] Evidence: [evidence]

A practical ad workflow is: confirm offer facts, define audience and awareness, generate angles, draft variants, review claims and platform requirements, select promising versions, then test them. ChatGPT can help with the first few steps; it cannot tell you how your audience will respond without actual test data.

ChatGPT prompts for email copywriting#

Email copy varies by relationship and purpose. A welcome email, a cold outreach message, and an abandoned-cart reminder should not all sound like the same sales pitch. Define the job of each message and keep the action clear.

34. Build a multi-email sequence

Draft a [number]-email sequence for [goal] and [audience]. First define the job of each email and how the sequence progresses. Use only this offer and proof: [data]. Keep each email focused on one primary action and vary the angle rather than repeating the same message. Flag missing facts instead of inventing them.

35. Write a welcome email

Draft a welcome email for new [subscribers/customers] who signed up for [reason]. Thank them, set expectations, and give them one useful next step. Use this brand voice: [voice]. Do not add benefits, content, or promises that aren’t confirmed here: [details].

36. Create a nurture email

Write a nurture email for [audience] who are considering [offer]. Its job is to [educate/address objection/share approved resource]. Use this customer insight: [research] and these verified facts: [facts]. Keep the CTA aligned with the reader’s likely next step.

37. Draft a product-launch email

Draft a product-launch email for [product] to [audience]. Include what is new, who it is for, why it may be relevant, and one CTA. Use only the launch details below. Do not imply limited availability or early-access benefits unless those conditions are confirmed.

38. Write a promotional email

Write a promotional email for [offer] to [audience]. The confirmed terms are [terms]. State the offer clearly, explain the most relevant benefit, and include one CTA. Do not add urgency, exclusions, savings, or deadlines beyond the terms provided.

39. Create a re-engagement email

Draft a re-engagement email for subscribers who have not [action] recently. Keep the tone respectful and non-judgmental. Offer one relevant next step based on [approved content or offer]. Do not claim to know why the subscriber disengaged.

40. Personalize cold outreach responsibly

Draft a concise cold outreach email to [role or organization type] about [relevant offer]. Personalization may use only these verified public details: [details and source]. Make the reason for contacting them clear, avoid pretending there is a prior relationship, and use one low-pressure CTA. Do not invent personal facts or results.

41. Write an abandoned-cart email

Draft an abandoned-cart email for [product] that reminds the recipient what they left in their cart and offers a helpful next step. Use only confirmed product and policy information: [details]. Do not assume the reason for the abandoned cart or add a discount, stock warning, or deadline unless supplied.

42. Plan a follow-up sequence

Create a [number]-message follow-up sequence after [event, such as demo or consultation]. Define the purpose of each message and the appropriate spacing only if timing guidance is provided. Keep the tone [tone]. Do not imply the recipient agreed to anything not stated in the source notes.

ChatGPT prompts for CTAs#

A call to action should fit the reader’s next step and the promise immediately before it. “Buy now,” “Book a demo,” and “Get the guide” ask for different levels of commitment. Provide the actual action, audience, and context.

43. Generate CTA options by intent

Write 12 CTA options for [offer], grouped by intent: low-friction next step, free trial, demo, purchase, lead magnet, consultation, and follow-up action. Only include options that correspond to actions actually available: [available actions]. Keep each CTA consistent with the copy immediately before it: [preceding copy].

44. Match CTA wording to funnel stage

Suggest five CTA options for people at the [awareness or funnel stage] stage. The next step is [action], and the page promise is [promise]. Keep the CTA clear about what happens after a click. Do not imply a free trial, instant access, or other condition unless confirmed.

ChatGPT prompts for brand voice#

A useful brand voice guide describes patterns you can observe in approved writing. It should not invent a brand personality from a few words or infer values that aren’t evident from the samples.

45. Build a voice guide from real samples

Analyze these approved brand samples. Build a voice guide covering tone, sentence rhythm, vocabulary, recurring patterns, phrases to avoid, and five before-and-after examples. Separate observable writing patterns from assumptions. Do not infer brand values that are not evident in the samples.

Approved samples: [paste examples]

46. Apply the voice guide to a draft

Rewrite the copy below to follow this brand voice guide. Preserve the factual meaning, offer, and CTA. After the rewrite, list the specific voice rules you applied and flag any part that conflicts with the guide. Do not add claims or details.

Voice guide: [guide] Copy: [draft]

47. Compare a draft with approved examples

Compare this draft with the approved brand samples. Identify differences in tone, vocabulary, sentence rhythm, and recurring patterns. Give specific examples from both the draft and samples. Recommend changes without altering factual claims.

Samples: [examples] Draft: [copy]

ChatGPT prompts for rewriting and editing copy#

Editing prompts work best when they specify what to improve and what must stay unchanged. “Make this more persuasive” is vague and can tempt the model to exaggerate. Ask it to diagnose the copy first, then revise within factual limits.

48. Edit for clarity and persuasion

Edit this copy for clarity and persuasion without changing factual meaning. First identify the five biggest weaknesses. Then rewrite the copy. Flag any claim that needs evidence instead of strengthening it.

Copy: [draft] Audience and goal: [details]

49. Make copy more concise

Reduce this copy by approximately [percentage or word count] while preserving the main message, essential details, and CTA. Remove repetition and filler, not necessary qualifications. List any content you removed that could change the meaning.

Copy: [draft]

50. Replace jargon with plain language

Identify technical terms, jargon, or abstract phrases in this copy. Suggest plain-language alternatives for [audience]. Preserve technical accuracy and flag any term that needs a subject-matter expert’s review. Do not simplify away important qualifications.

Copy: [draft]

51. Make benefits more specific

Review this copy for vague benefits. For each one, explain what information would make it more concrete. Rewrite only benefits supported by the facts below; leave unsupported ones flagged rather than inventing details.

Copy: [draft] Verified facts: [facts]

52. Adapt the tone

Rewrite this copy in a [tone] tone for [audience]. Preserve the factual meaning, offer, and CTA. Do not use humor, slang, urgency, or emotional intensity unless it fits the tone and is supported by the brief.

Copy: [draft] Tone guidance: [rules]

53. Remove weak or unsupported claims

Review this copy for vague, exaggerated, or unsupported claims. Categorize each as: supported by supplied evidence, needs more evidence, or unclear. Suggest a narrower alternative only when the supplied facts support it. Do not certify legal or platform compliance.

Copy: [draft] Evidence: [sources]

ChatGPT prompts for A/B test ideas#

AI can help you generate test hypotheses, but it cannot guarantee a winning variation. A test should compare one meaningful variable at a time—such as the promise, proof, CTA, objection, framing, or page sequence—so that the results can inform a decision.

54. Generate testable hypotheses

Based on this page and customer evidence, propose eight A/B test hypotheses. For each, include the variable, hypothesis, customer insight, expected behavioral effect, and metric that could indicate improvement. Change one meaningful variable per test. Do not predict a guaranteed winner.

Page: [page or section] Customer evidence: [research] Goal and primary metric: [goal and metric]

55. Prioritize test ideas

Rank these A/B test ideas by clarity of hypothesis, relevance to the conversion goal, strength of supporting evidence, and ease of implementation. Explain your ranking and identify what data is missing. Do not treat the ranking as a prediction of results.

Test ideas: [ideas] Context: [goal, audience, page]

56. Check whether a test changes too much

Review this proposed A/B test. Identify every variable that changes between version A and version B. Recommend a simpler comparison that tests one meaningful idea while preserving necessary legal, factual, and user-experience information.

Version A: [copy] Version B: [copy]

How copywriters should review AI-generated copy#

AI-generated copy is a draft, not a source of truth. Before publishing, check it against the brief, the product, the evidence, and the intended audience. A quick review can catch errors that sound polished enough to escape a casual read.

For claims about results, ask: “What evidence supports this exact wording?” For testimonials, confirm that the person and statement are real and that any required permissions or disclosures are handled. For product claims, check current documentation with the product or subject-matter team.

Do not ask ChatGPT to invent testimonials, reviews, statistics, certifications, pricing, scarcity, guarantees, or customer results. Do not treat the model’s own claim-check as a legal review or compliance approval. Google Ads prohibits false, misleading, or unrealistic claims, but advertisers still need to review current policies and the specific context of their ads.

How to use multiple AI models for copywriting#

Different AI models may interpret the same brief in different ways. Comparing their outputs can help you find alternative angles, spot assumptions, and identify weak parts of a draft. It doesn’t mean that the most popular answer is correct—or that several models agreeing proves a claim is true.

A practical workflow is to give each model a distinct job:

1. Model A: Research synthesis and message strategy. Provide the same source material and ask it to extract audience language, objections, and evidence. Require it to separate direct evidence from interpretation.
2. Model B: Alternative angles. Ask for distinct approaches to the same goal—such as benefit-led, problem-aware, or proof-led messaging—using the same approved facts.
3. Model C: Critical editing. Ask it to review a chosen draft for clarity, unsupported claims, tone mismatches, and missing information.

Compare the points where models disagree. Is one assuming a different audience awareness level? Has one turned a feature into an unsupported outcome? Does one model lean on a stronger promise than the evidence allows? These disagreements are useful review signals, not a vote on the right answer.

UltraGPT can help copywriters compare outputs from multiple AI models in one place. Use it to explore different messaging angles, challenge weak wording, and refine drafts without switching between platforms. Keep the human copywriter responsible for final claims, voice, positioning, and conversion decisions.

Compare copy ideas from multiple AI models in one place. Explore different angles, challenge weak messaging, and refine stronger copy without switching between platforms.

Practical copywriting prompt template#

Use this master prompt when you need a structured start. It asks the model to identify gaps before drafting and keeps the copywriter in control of the direction.

You are helping me create [asset] for [audience].

Offer: [offer] Goal: [goal] Audience pains and desires: [research or customer language] Awareness level: [stage] Verified proof: [proof] Brand voice: [voice] Constraints: [length, format, channel rules, required wording, prohibited claims]

First, identify missing information that could materially affect the copy. Do not guess. Then propose three distinct messaging angles, explaining what customer insight each one uses and what evidence it requires. Wait for me to choose an angle. After I choose, draft the copy in [format and length]. Do not invent facts, proof, testimonials, product capabilities, customer results, pricing, guarantees, certifications, or scarcity.

After the first draft, continue with focused instructions. For example:

Check the draft against the original brief. List any unsupported claims, missed constraints, unclear phrases, and points where the CTA does not match the promise. Do not rewrite the copy yet.

Then, once you’ve reviewed the feedback:

Revise the draft to address the approved issues. Preserve the factual meaning and use only the supplied evidence. Flag anything that still needs human confirmation.

Keep the copywriter in control#

The most useful copywriting prompts don’t ask ChatGPT to “write something persuasive” and hope for the best. They supply a real brief, give the model clear boundaries, and create a process for checking what comes back. Use prompts to explore ideas and speed up drafts, then apply your expertise to the claims, voice, and decisions that matter.

Ready to improve your copywriting workflow? Use multiple AI models with UltraGPT to compare messaging angles, refine drafts, and choose stronger copy faster.