Everyone is using ChatGPT for marketing. Most of them are using it badly. They type "write me a marketing email" and get back something that reads like it was written by a committee that has never purchased anything online. Then they declare AI useless for marketing and go back to writing everything from scratch.
The problem is not ChatGPT. The problem is how marketers use it. The difference between a mediocre AI-assisted marketer and a highly productive one comes down to three things: prompt specificity, workflow design, and knowing when to use AI versus when to think for yourself.
This guide gives you the exact prompts, workflows, and use cases that produce real marketing output -- not the generic "act as a marketing expert" advice you find everywhere else. Everything here comes from using ChatGPT daily for marketing work across SaaS, e-commerce, and services businesses.
The Prompting Framework That Actually Works
Forget the "act as" prompt templates. They produce generic output because they give ChatGPT generic instructions. Marketing is specific by nature -- specific audience, specific offer, specific context. Your prompts need to match.
The SCOPE Framework
Every marketing prompt should include five elements:
S - Situation: What is the context? What product, what stage of the customer journey, what has happened so far?
C - Constraint: What are the boundaries? Word count, tone, format, what to avoid?
O - Outcome: What should the reader do after engaging with this content?
P - Persona: Who is the audience? Be specific about demographics, psychographics, and current state of mind.
E - Examples: What does good look like? Include samples of your brand voice, competitor examples, or formats you like.
A weak prompt:
Write an email to promote our new feature.
A SCOPE prompt:
Situation: We just launched a dashboard redesign for our project management tool. Users have been requesting this for 8 months. The update includes customizable widgets, dark mode, and 40% faster load times. / Constraint: 120-150 words. Conversational tone, not corporate. No exclamation marks. One CTA only. / Outcome: Get users to log in and try the new dashboard today. / Persona: Active users who log in at least 3x per week. They are project managers at companies with 20-100 employees. They are busy and skeptical of "new and improved" claims. / Example of our voice: "We rebuilt the dashboard because yours was slow. Here is what changed."
The second prompt produces copy you can actually use. The first produces copy you throw away.
Specific Prompts for Every Marketing Function
Here are the prompts I use regularly, organized by function. Copy them, replace the bracketed sections with your details, and iterate from the output.
Email Marketing Prompts
Welcome email:
Write a welcome email for [product/service]. The subscriber just signed up for [free trial / newsletter / lead magnet]. Key information to convey: [3 main points]. Tone: [warm but not gushy / professional / casual]. Length: [100-150 words]. End with a single action the reader should take next. Do not use the phrase "we are excited" or "welcome aboard."
Cold outreach:
Write a cold outreach email to [job title] at [company type]. I am reaching out because [specific reason tied to their situation]. My product solves [specific problem]. The email must be under 100 words, open with a line that references something specific about their company or role, and end with a low-commitment CTA like booking a 15-minute call or watching a 2-minute demo. No corporate buzzwords. No "I hope this email finds you well."
Cart abandonment:
Write a cart abandonment email for [product type] at [price point]. The customer added the item [timeframe] ago. The email should acknowledge they are busy, not guilt-trip them. Include one reason to complete the purchase that is not a discount. Tone: helpful, not desperate. Under 100 words.
Re-engagement:
Write a re-engagement email for subscribers who have not opened an email in [timeframe]. Acknowledge the silence honestly. Give them one compelling reason to stay subscribed. Include an easy one-click unsubscribe option framed positively. Tone: respectful of their time. Under 120 words.
Social Media Prompts
LinkedIn post:
Write a LinkedIn post about [topic]. Target audience: [job titles/industry]. The post should open with a specific, counterintuitive observation -- not a generic statement. Structure: hook (1-2 lines), story or example (3-4 short paragraphs), insight or lesson (2-3 lines), question to drive comments. Length: 150-200 words. Do not use "Here is the thing:" or "Let me tell you why" or "Thoughts?" at the end.
X/Twitter thread:
Write a 7-tweet thread about [topic] for [audience]. First tweet must be a standalone hook that makes people want to read the rest. Each subsequent tweet should contain one complete idea. Use specific numbers, examples, or contrasts -- not abstract advice. Final tweet should summarize the key takeaway and include a soft CTA. No hashtags. No emojis. Maximum 240 characters per tweet.
Instagram caption:
Write an Instagram caption for a post showing [describe the image/video]. Target audience: [description]. The caption should start with a hook that stops the scroll -- a question, surprising fact, or bold statement. Keep it under 125 words. Include a CTA that drives comments, not just likes. Suggest 5-8 relevant hashtags at the end, mixing popular (100K+ posts) with niche (10K-100K posts).
Ad Copy Prompts
Facebook/Meta ad:
Write 5 variations of primary text for a Meta ad promoting [product/service]. Target audience: [description]. The ad should address [specific pain point]. Each variation should use a different angle: social proof, curiosity, direct benefit, fear of missing out, and contrarian take. Keep each under 125 characters for primary text. Include a suggested headline (under 40 characters) and description (under 30 characters) for each.
Google Search ad:
Write 5 responsive search ad headline combinations for the keyword [keyword]. Each headline must be under 30 characters. Include the keyword naturally in at least 2 headlines per combination. Write 3 description lines under 90 characters each. Focus on [unique selling proposition]. Include specific numbers where possible (percentages, dollar amounts, time saved).
Customer Research Prompts
Persona building:
Based on the following customer reviews and testimonials, create a detailed buyer persona. Include: demographics, job title and responsibilities, goals, frustrations, objections to purchasing, preferred content formats, and the specific language they use to describe their problems. Here are the reviews: [paste 10-20 reviews].
Competitor analysis:
Analyze the marketing approach of [competitor name] based on their website copy at [URL description], their social media presence, and their positioning. Identify: their primary target audience, their key value propositions, their messaging tone, the objections they address, and gaps in their messaging that a competitor could exploit. Format as a brief with actionable takeaways.
Review mining:
Analyze these [product category] reviews from [source]. Extract: the top 5 complaints, the top 5 praised features, the exact language customers use to describe their problems (not paraphrased), and the emotional triggers that appear most frequently. Organize by frequency. Here are the reviews: [paste reviews].
Blog and Content Prompts
Blog outline:
Create a detailed outline for a blog post titled [title]. Target keyword: [keyword]. Target audience: [description]. The outline should include: H2 sections with brief descriptions of what each covers, H3 subsections where needed, suggested word count per section, and a notes section for each H2 explaining the specific value it provides to the reader. Total target length: [word count]. The post should be structured to answer the search intent behind the keyword, not just cover the topic generally.
Content repurposing:
Take this blog post and create: 3 LinkedIn posts (each highlighting a different key point), 5 tweets that can work standalone, 1 email newsletter summary (150 words), and 3 Instagram caption ideas. Maintain the original tone and key insights but adapt the format and length for each platform. Here is the post: [paste blog post].
Workflow Integrations That Save Real Time
Prompts alone are not a workflow. The real productivity gains come from integrating ChatGPT into your existing marketing processes.
The Content Pipeline Workflow
Step 1: Research phase. Use ChatGPT to analyze competitor content, mine customer reviews for language and pain points, and generate content angles. This replaces 2-3 hours of manual research with a 30-minute AI-assisted session.
Step 2: Brief creation. Feed the research output back to ChatGPT and ask it to create a content brief with outline, key points, and target audience notes. Review and adjust the brief -- this is where your strategic input matters most.
Step 3: First draft. Use the brief to generate a first draft. For blog posts, generate section by section rather than asking for the full post at once. This produces better output because each section gets focused attention.
Step 4: Human editing. This is non-negotiable. Edit the draft for voice, accuracy, specific examples from your experience, and any claims that need verification. Budget 30-40 percent of the total writing time for editing.
Step 5: Repurpose. Feed the final post back to ChatGPT and generate social media posts, email snippets, and ad copy from it. One piece of content becomes ten.
The Email Campaign Workflow
Weekly newsletter process:
- Feed ChatGPT your top 3-5 content pieces or news items for the week
- Ask it to generate a newsletter draft with brief summaries, transitions, and a consistent voice
- Generate 5 subject line options
- Edit the draft and pick the subject line (or A/B test the top two)
- Generate preview text and social sharing copy from the same session
This takes a weekly newsletter from a 3-hour task to a 45-minute task.
The Ad Testing Workflow
Generating ad variations at scale:
- Write your best-performing ad copy manually
- Feed it to ChatGPT: "Here is my best-performing ad for [product]. Generate 10 variations that test different angles: emotional appeal, rational benefits, social proof, scarcity, and curiosity. Maintain the same offer and CTA."
- Review the variations and select 4-5 for testing
- After testing, feed the performance data back: "Here are the results. The top performer used [angle]. Generate 5 more variations that lean into this angle while testing different hooks."
This iteration cycle compresses weeks of manual creative testing into days.
Custom GPTs for Marketing
Custom GPTs are one of the most underused ChatGPT features for marketing. Instead of re-entering your brand guidelines, tone preferences, and audience details every session, you build a Custom GPT that knows all of this permanently.
Brand Voice GPT
Configure a Custom GPT with:
- Your brand guidelines document
- 10-20 examples of your best content (emails, social posts, blog excerpts)
- A list of words and phrases you never use
- Your target audience description
- Your product/service description
Now every content request starts with context that would take 500+ words of prompting to reproduce manually.
Customer Research GPT
Configure with:
- Your customer review database (or representative samples)
- Your buyer personas
- Competitor information
- Industry terminology and jargon your customers use
Use this GPT for ongoing review analysis, persona refinement, and message testing. It gets better as you feed it more data over time.
Campaign Planning GPT
Configure with:
- Your marketing calendar and recurring campaigns
- Historical campaign performance data
- Your channel strategy and posting frequency
- Budget constraints and KPI targets
This becomes your campaign planning assistant that understands your context without re-explanation.
ChatGPT vs. Claude for Marketing: An Honest Comparison
I use both daily. Here is where each one wins.
ChatGPT is better for:
Speed and iteration. When you need 20 subject line variations or 10 ad headlines fast, ChatGPT's response speed and willingness to produce large lists without pushback is an advantage.
Brainstorming volume. Ask for 50 content ideas and ChatGPT delivers 50 usable options. Claude tends to deliver fewer but more considered suggestions.
Short-form copy. Tweets, ad headlines, email subject lines, CTAs -- ChatGPT handles these bite-sized copy tasks efficiently.
Integrations. ChatGPT's plugin ecosystem, Custom GPTs, and API integrations connect it to more marketing tools than Claude's current ecosystem.
Claude is better for:
Long-form content. Blog posts, whitepapers, case studies, and guides come out more cohesive and better structured from Claude. The output requires less structural editing.
Brand voice consistency. Claude maintains a consistent voice across a long document better than ChatGPT, which tends to drift in tone over longer outputs.
Complex instructions. When your prompt has multiple constraints and requirements, Claude follows them more precisely. ChatGPT sometimes ignores or reinterprets specific instructions.
Nuanced analysis. Competitive positioning, strategic frameworks, and analytical tasks that require balanced reasoning produce stronger output from Claude.
Honest limitations. Claude is more likely to tell you when it cannot do something well or when your approach has a flaw. ChatGPT tends to give you what you ask for even when the ask is misguided.
The practical recommendation:
Use ChatGPT Plus ($20/month) for daily quick tasks -- social media drafts, email subject lines, ad variations, brainstorming sessions. Use Claude Pro ($20/month) for weekly strategic work -- blog content, strategy documents, competitive analysis, and any content over 500 words. The $40/month combined investment pays for itself in the first week through time savings.
If you can only choose one, Claude produces better marketing content that requires less editing. ChatGPT is more versatile for the quick, repetitive tasks.
Limitations You Need to Know
ChatGPT does not know your customers. It knows general patterns about marketing. The difference matters. Every piece of output needs to be filtered through your specific knowledge of what your audience actually responds to.
Factual accuracy. ChatGPT generates plausible-sounding statistics and claims that may not be accurate. Never publish statistics, case studies, or data points from ChatGPT without verification. This is especially important for ad copy, where false claims can trigger platform policy violations or legal issues.
Brand differentiation. If you use ChatGPT the same way everyone else does, your content will sound like everyone else's. The competitive advantage comes from the human layer -- your experience, your opinions, your specific examples. AI is the production tool. You are the differentiation.
Recency. ChatGPT's training data has a cutoff. For marketing tasks that depend on current trends, platform algorithm changes, or recent industry developments, verify the information against current sources.
Legal and compliance. AI-generated marketing content can inadvertently make claims that violate FTC guidelines, platform advertising policies, or industry regulations. Always have a human review for compliance, especially for healthcare, finance, and legal marketing.
The overproduction trap. ChatGPT makes it easy to produce more content than your audience wants. Sending three emails per week instead of one because "AI makes it easy" does not improve your marketing -- it increases unsubscribes. The constraint on content should be audience appetite, not production capacity.
Getting Started This Week
Do not try to implement everything at once. Start with the highest-leverage use case for your situation.
If you write marketing emails: Use the email prompts above to draft your next three emails. Compare the AI-assisted time against your usual process. Measure the time savings.
If you manage social media: Use the batch content creation workflow. Generate a week of posts in one session. Edit them. Schedule them. Track engagement against your manually-created content baseline.
If you run ads: Take your current best-performing ad and generate 10 variations using the ad testing workflow. Run them as a test. Let data pick the winner.
If you create long-form content: Use the content pipeline workflow for your next blog post or newsletter. Focus on the research and outline phases -- this is where AI saves the most time with the highest quality output.
The marketers who get the most from ChatGPT are not the ones with the fanciest prompts. They are the ones who know their audience deeply, use AI as a production accelerator, and never publish anything without applying their own judgment. The tool is fast. You need to be thoughtful. That combination is what makes AI-assisted marketing work.