ChatGPT Prompts for Marketing Focus (That Actually Work)

Why Most Marketers Get Terrible Output from ChatGPT

I've watched agency owners and founders spend 45 minutes wrestling with ChatGPT and end up with something that sounds like it was written by a committee of interns. The problem is almost never the tool. It's the prompt.

Generic input produces generic output. If you type "write me a marketing campaign for my SaaS," you're going to get a generic five-bullet response that fits every SaaS company and none of them. The fix is specificity: role assignment, clear constraints, context, and output format. Once I started engineering prompts the right way across my own companies, the output quality jumped dramatically.

And I'm not alone in this. 65% of marketers are now using AI regularly, and the top three ways they use it are data analysis, generating written content, and personalization. The tool has hit mass adoption. The differentiator is no longer whether you use it - it's how well you prompt it.

This article breaks down ChatGPT prompts for marketing focus organized by use case: strategy, lead generation, cold outreach, content, email marketing, SEO, analytics, and ad copy. I'll give you the exact prompt structures and explain why they work - because understanding the mechanics means you can adapt them instead of just copying and pasting.

The Anatomy of a High-Output Marketing Prompt

Before diving into specific prompts, understand this framework. Every high-performing ChatGPT prompt for marketing has four components:

• Role: Tell ChatGPT who it is. "You are a direct-response copywriter with 10 years of B2B SaaS experience" produces better output than just asking a question cold.
• Context: Give it your business, audience, offer, and any constraints. The more specific your context, the more specific the output.
• Task: State exactly what you want produced - not a vague request, but a deliverable. "Write three email subject line variations" beats "help me with emails."
• Format: Tell it how to structure the output. "Give me the output in a numbered list with the subject line, preview text, and a one-line rationale for each" saves you massive editing time.

Keep that framework in mind for every prompt below. You'll see it embedded in each one.

One more thing before we go section by section: I recommend creating a "context block" you paste at the start of any new ChatGPT session. It should contain your ICP, your core offer, your differentiators, your brand voice guidelines, and any banned phrases. Paste it before the first prompt every time. This eliminates 80% of the back-and-forth you'd otherwise spend re-explaining your business.

Marketing Strategy Prompts

ICP Identification Prompt

Most marketers skip this step and wonder why their campaigns don't convert. Use this before writing a single word of copy:

"You are a B2B marketing strategist. I sell [product/service] to [general market]. Based on the highest-converting customer profiles in similar markets, identify three distinct ICP segments. For each segment, describe: job title, company size, primary pain point, how they currently solve the problem, and the emotional trigger that would make them take action now. Format as a numbered list."

Run this before you build any campaign. Feed the output back into every subsequent prompt as context. This alone eliminates the generic problem.

Positioning Statement Prompt

"You are a positioning strategist. My product is [product]. My top three competitors are [A, B, C]. My strongest customers are [ICP]. Write five positioning statement options that differentiate us from competitors on [speed / price / ease of use / specific outcome]. Each statement should be one sentence, jargon-free, and focused on a specific customer outcome - not features."

This is particularly useful when you're launching a new offer or entering a market where three incumbents already have brand recognition. You're not going to out-budget them - you're going to out-position them.

Campaign Brief Prompt

"You are a performance marketing director. I need a campaign brief for [goal: leads / demos / free trial signups] targeting [ICP]. My offer is [offer]. My budget is [budget range]. The campaign will run on [channels]. Write a brief that includes: campaign goal, target audience definition, core message, channel strategy, three ad concepts, and KPIs to track. Be specific and opinionated - don't hedge."

The "be specific and opinionated" instruction at the end matters more than it looks. Without it, ChatGPT tends to give you balanced, wishy-washy recommendations. Telling it to take a stance forces a more useful output.

Competitive Analysis Prompt

This is one of the most underused prompt categories in marketing. Before you build a campaign, you need to understand what your competitors are doing - and where they're weak.

"You are a competitive intelligence analyst. I need a competitor breakdown for [your product category]. My top three competitors are [A, B, C]. For each competitor, analyze: (1) their primary value proposition based on their marketing, (2) the audience they appear to be targeting, (3) their apparent weaknesses or gaps in their messaging, (4) what a prospect would need to believe to choose them over me. Then identify three positioning gaps I could exploit. Be direct."

Pair this with real market data. ChatGPT's training data has gaps, so validate any competitor claims you plan to act on. For more prompts built specifically around market research and competitive intelligence, grab my free GPT Market Research Prompts - I built that resource specifically for founders and agency owners who need to understand a market fast.

Customer Journey Mapping Prompt

"You are a customer experience strategist. Map out the full customer journey for [ICP] buying [product/service]. Cover these stages: pre-awareness, awareness, consideration, evaluation, purchase, and post-purchase. For each stage, describe: what the customer is thinking and feeling, what information they need, what channel they're likely on, and what the ideal touchpoint from us looks like. Output as a table."

This gives you a complete content and campaign roadmap in one prompt. Each stage of the journey becomes a brief for a different asset - ad, email, landing page, case study, onboarding sequence.

Lead Generation Prompts

Prospect List Strategy Prompt

Before you can market to anyone, you need a list. Here's a prompt that helps you figure out exactly where to find your best prospects:

"You are a B2B lead generation expert. My ICP is [title] at [company type] companies with [employee range] employees in [industry]. List 10 specific places or sources where I can find concentrated lists of these prospects. For each source, explain why it's relevant and how I'd use it to build a list. Prioritize sources that offer direct contact information or can be scraped."

Once ChatGPT gives you those sources, you need to actually pull the data. For LinkedIn-sourced prospects, ScraperCity's B2B email database lets you filter by title, industry, company size, and seniority to pull verified contact data at scale. The AI tells you where to look; the scraper does the heavy lifting of actually getting the contacts.

Lead Magnet Prompt

"You are a conversion optimization specialist focused on B2B lead generation. My audience is [ICP]. My core offer is [offer]. Create five lead magnet concepts that would attract my ideal buyer at the top of funnel - before they're ready to buy. For each concept, specify: the format (checklist, template, calculator, etc.), the exact title, the core promise, and why someone in my ICP would trade their email address for it."

The best lead magnets solve one small, specific problem. ChatGPT is excellent at generating variations of that once you give it enough context about who your buyer is.

Webinar Topic and Promotion Prompt

Webinars are one of the highest-converting lead gen formats for B2B. Here's how to use GPT to build an entire webinar campaign in under an hour:

"You are a B2B demand generation specialist. I want to run a webinar to generate [X] leads among [ICP]. My expertise is in [topic area]. Suggest five webinar topic angles that would get a [title] to register. For the strongest angle, write: a registration page headline, three bullet points that describe what attendees will learn, a 50-word email invitation, and three subject line options for the invite. Keep every word benefit-focused."

For more structured lead gen prompt sequences, check out my GPT Lead Gen Prompts resource - it's a free download with prompts organized specifically around the prospecting workflow.

Re-Engagement Campaign Prompt

Your existing list is your most underutilized asset. Most founders focus entirely on net-new leads and ignore the warm contacts sitting in their CRM going cold.

"You are an email marketing strategist. I have a list of [X] contacts who engaged with us [timeframe] ago but have not responded since. I need a re-engagement campaign. Write a five-email sequence to win back inactive prospects. Each email should: use a different hook (curiosity, new information, social proof, a direct question, a value offer), be under 100 words, and have a clear low-friction CTA. The goal is a reply, not a sale. Avoid phrases like 'just checking in.'"

The low-friction CTA instruction is critical. Asking a cold contact to book a 30-minute call is too much friction. Ask them a yes/no question, give them a useful piece of content, or offer a quick win before you ask for their time.

Cold Outreach Prompts

This is where I've spent the most time personally. I've written cold emails that generated millions in pipeline, and I've used GPT to scale what works.

Cold Email Framework Prompt

"You are a direct-response cold email copywriter. My prospect is a [title] at a [company type] company. They are experiencing [specific pain point]. My solution delivers [specific outcome] in [timeframe]. Write three cold email variations using these frameworks: (1) Problem-Agitate-Solve, (2) Case Study Lead, (3) Curious Question Lead. Each email must be under 100 words, have no more than one CTA, and avoid phrases like 'I hope this finds you well' or 'I wanted to reach out.' Format: subject line, preview text, email body."

The constraint on word count and banned phrases is critical. Without it, ChatGPT writes emails that sound exactly like every AI-generated email in every inbox right now. You want it to produce something that reads like a human wrote it.

Follow-Up Sequence Prompt

"You are a cold outreach specialist. I have a four-email follow-up sequence. Email 1 was [paste email 1]. Write three follow-up emails spaced across 14 days. Each follow-up should: add a new angle or piece of value (not just 'checking in'), stay under 75 words, and have a different CTA than the previous email. The sequence goal is [demo booking / free trial / reply to qualify]. Avoid repetition across emails."

If you're running cold outreach at scale, pair these prompts with a sending platform like Smartlead or Instantly - both handle deliverability and sequence automation so your AI-generated copy actually reaches inboxes.

LinkedIn Outreach Message Prompt

Cold email gets a lot of attention, but LinkedIn DMs convert well for certain ICPs - especially enterprise buyers who don't respond to cold email at all.

"You are a LinkedIn outreach specialist focused on B2B sales. My prospect is a [title] at a [company type] company. I noticed [specific thing about their profile, recent post, or company news]. My offer delivers [specific outcome] for [ICP]. Write three LinkedIn connection request messages and two follow-up DM sequences. Each connection request must be under 300 characters. Each DM must be conversational, not salesy, and under 75 words. No pitching in the first message."

The "no pitching in the first message" rule is not optional. LinkedIn penalizes accounts that spam connection requests, and even more importantly, prospects don't accept connections from people who lead with a pitch. Build the connection first.

I go much deeper on structuring cold email campaigns inside Galadon Gold - including how to combine prompt-generated copy with real personalization at scale.

Personalization at Scale Prompt

The hardest part of outreach is making 500 emails feel like they were each written individually. Here's a prompt structure that gives you a personalization formula you can apply systematically:

"You are a cold email personalization expert. I am sending outreach to [ICP]. I have access to the following data points for each prospect: [LinkedIn headline, company industry, company size, recent funding news, recent LinkedIn post topic]. Write a personalization formula - a one to two sentence opening that uses one of these data points to show I did my homework. Give me five different template variants for this opening section. Each variant should use a different data point and a different tone (curious, empathetic, direct, complimentary, data-driven)."

Once you have the formula, feed your prospect data to a tool like Clay to apply personalizations at scale. The prompt gives you the template; the tools do the execution.

For more outreach-specific GPT prompts, grab the free Cold Email GPT Prompts pack - it's organized by sequence type and includes prompts for both initial outreach and follow-ups.

Content Marketing Prompts

Content Pillar Prompt

"You are a B2B content strategist. My brand is [brand]. My ICP is [ICP]. My core offer is [offer]. Identify five content pillars that would attract my ICP across the buyer journey - from awareness to decision. For each pillar: name it, describe the types of content that belong under it, give three specific topic examples, and explain which stage of the buyer journey it addresses."

This is how you build a 90-day content calendar in an afternoon instead of a week.

Blog Post Brief Prompt

"You are an SEO content strategist. I need a detailed brief for a blog post targeting the keyword '[keyword].' My audience is [ICP]. Write a brief that includes: the search intent behind the keyword, a recommended H1 title, an outline with H2 and H3 headings, the primary question the post must answer, three supporting statistics or data points to research, internal linking opportunities, and recommended CTA. Do not write the post itself - just the brief."

The instruction "do not write the post itself" sounds counterintuitive, but getting the brief right first produces dramatically better drafts when you then ask it to write the article. It forces the thinking before the execution.

Content Repurposing Prompt

Most marketers create one piece of content and publish it once. That's leaving enormous reach on the table. One good insight can become a blog post, a LinkedIn post, three tweets, an email, a YouTube script, and a short-form video - if you have the right prompts to transform it.

"You are a content repurposing specialist. I have the following piece of content: [paste content]. Repurpose it into the following six formats: (1) a 1,500-word SEO blog post outline, (2) a LinkedIn post under 300 words with a hook, insight, and CTA, (3) three Twitter/X threads of 5-7 tweets each with different angles, (4) a 5-minute YouTube video script with sections, (5) a 200-word email newsletter intro, (6) five Instagram caption options under 150 characters. Maintain the core insight across all formats but rewrite each for the platform norms."

This single prompt turns one interview, podcast episode, or strong insight into a week's worth of content across every channel your audience is on.

LinkedIn Post Prompt

"You are a B2B LinkedIn content creator. My audience is [ICP]. Write five LinkedIn post hooks on the topic of [topic] that would stop a [title] from scrolling. Each hook should use a different structure: a counterintuitive claim, a specific number, a relatable pain point, a short story opener, and a provocative question. No emojis. No filler. No corporate speak."

LinkedIn rewards specificity and personality. The "no corporate speak" instruction is non-negotiable if you want engagement.

Video Script Prompt

Short-form video is the highest-reach content format right now for most B2B audiences on LinkedIn and YouTube. Here's how to get ChatGPT to write scripts that actually hold attention:

"You are a short-form video scriptwriter for B2B audiences. Write a 60-second video script on the topic of [topic] for a [title] audience. The script should follow this structure: (1) a hook in the first five seconds that states a surprising claim or problem, (2) a middle section that delivers the core insight or three-step framework, (3) a close with a specific CTA. Write it as spoken word - not bullet points, not headers. Use plain language. No jargon. Target reading pace: 150 words per minute."

The reading pace instruction matters for video because it ensures the script fits the time constraint without feeling rushed. 150 words per minute is a natural conversational pace for on-camera delivery.

Email Marketing Prompts

Email remains one of the highest-ROI marketing channels. But most marketers treat their list as a broadcast channel instead of a conversation. These prompts are built to change that.

Welcome Sequence Prompt

Your welcome sequence is the most important email sequence you'll ever write. New subscribers are most engaged in the first 72 hours. If you blow this window, you've already lost the relationship.

"You are an email marketing specialist. Write a five-email welcome sequence for new subscribers to [brand/offer]. The subscriber signed up because of [lead magnet or opt-in offer]. The sequence goal is to build trust, deliver immediate value, and move subscribers toward [desired action: booking a call / starting a trial / buying a product]. Email 1: immediate welcome and delivery of the lead magnet. Email 2: your backstory and why this matters. Email 3: one actionable tip or insight. Email 4: social proof or case study. Email 5: soft CTA. Each email under 200 words. Conversational tone."

Newsletter Subject Line Prompt

Most email subject line advice is generic. Here's a prompt that forces ChatGPT to produce genuinely differentiated options:

"You are a direct-response email copywriter. Write 15 subject line options for a newsletter about [topic]. Use these specific angles across the 15 options: curiosity gap (3), specific number or list (3), controversial opinion (2), personal story hook (2), pain point (2), and benefit-led (3). Every subject line must be under 50 characters. Flag the top three based on predicted open rate for a [ICP] audience."

Asking ChatGPT to flag its own top picks forces it to add a layer of analysis rather than just dumping 15 options with equal weighting. The flagging step usually reveals its best thinking.

Nurture Sequence Prompt

"You are an email marketing strategist. I need a 10-email nurture sequence for [ICP] who are in the consideration stage - they know about us but haven't bought yet. My core offer is [offer]. The sequence should cover: education (3 emails), social proof (2 emails), objection handling (2 emails), urgency/scarcity (1 email), and direct CTA (2 emails). For each email, specify the goal, the hook, the core message, and the CTA. Don't write the full emails - just the strategy for each."

Use this as your sequence architecture, then feed each email slot back into ChatGPT individually to write the full version. You'll get dramatically better results than asking it to write all 10 emails in one shot.

Re-Engagement Email Prompt

"You are an email deliverability expert and copywriter. I have a segment of [X] inactive subscribers who have not opened an email in [timeframe]. Before I suppress them, I want to run a three-email re-engagement campaign. Write three emails: (1) a curiosity-led 'are you still there?' email, (2) a value delivery email with one free resource, (3) a final 'we're removing you' email that creates FOMO. All three under 100 words. The goal is a click or reply, not a sale."

The final email in this sequence - the "we're removing you" format - consistently outperforms the other two in re-engagement campaigns I've run. Tell people you're about to remove them and watch how many suddenly click.

SEO and Search Prompts

ChatGPT is a surprisingly powerful SEO tool when you know how to use it. The key is combining its language capabilities with your own keyword research and search intent analysis.

Keyword Clustering Prompt

"You are an SEO strategist. I have the following list of keywords: [paste 20-30 keywords]. Group them into topical clusters based on shared search intent. For each cluster: name the cluster, list the keywords in it, identify the primary keyword, suggest a content format (pillar page, blog post, landing page, FAQ), and explain the searcher intent behind the cluster. Output as a table."

This replaces hours of manual keyword mapping. Paste in your keyword research output and let ChatGPT organize it into a content architecture you can execute against.

On-Page SEO Optimization Prompt

"You are an on-page SEO specialist. I have the following blog post: [paste post]. The target keyword is [keyword]. Review the post and provide specific recommendations for: H1 and H2 structure improvements, keyword density and placement issues, internal linking opportunities, meta title and description rewrites (under 60 and 155 characters respectively), any thin sections that need expansion, and a recommended FAQ section of five questions with answers that could target featured snippets. Be specific, not general."

FAQ and Featured Snippet Prompt

Featured snippets and People Also Ask boxes are among the highest-visibility real estate in search results. Here's how to target them systematically:

"You are an SEO content writer. For the keyword '[keyword],' generate 10 FAQ questions a searcher might have. For each question, write a 40-60 word answer formatted to win a featured snippet - direct, factual, structured, and using the keyword naturally in the first sentence. Format as a Q&A list ready to embed in a blog post."

Plug this into the bottom of any existing post. It takes 15 minutes and often doubles the amount of search traffic that post earns within 60 days of Google recrawling it.

Ad Copy Prompts

Facebook and Instagram Ad Prompt

"You are a paid social media copywriter specializing in direct response. My product is [product]. My target audience is [audience]. Write four ad variations for Facebook and Instagram: one using a testimonial-style hook, one using a pain-point lead, one using a bold claim, and one using a story format. For each ad: write the primary text (under 125 words), a headline (under 40 characters), and a description (under 30 characters). Include a clear CTA in every ad."

Google Search Ad Prompt

"You are a Google Ads copywriter. I am targeting the keyword '[keyword]' for [product/service]. My ICP is [ICP]. Write 10 headline variations (max 30 characters each) and 5 description variations (max 90 characters each) for a Responsive Search Ad. Headlines should cover: feature-led, benefit-led, problem-led, social proof-led, and urgency-led angles. Flag which headlines pair best together."

The character counts matter. If you don't specify them, ChatGPT will write headlines that get truncated in the actual ad - wasted output.

Retargeting Ad Prompt

Most ad accounts leave money on the table by running the same creative to cold and warm audiences. Retargeting audiences have already shown intent - they need a different message entirely.

"You are a paid media strategist. I need retargeting ad copy for website visitors who visited [specific page] but did not convert. My offer is [offer]. The visitor's objection is likely [price / trust / timing / not sure yet]. Write four retargeting ad variations that address each of these objections directly. Each variation: primary text under 100 words, headline under 40 characters. Tone should be confident, not desperate. Include social proof where possible."

Notice the "confident, not desperate" instruction. Retargeting ads often read like the brand is chasing the prospect. That framing hurts conversions. Confident copy that assumes the prospect is still interested - but needs one more piece of information - converts better.

YouTube Pre-Roll Ad Script Prompt

"You are a YouTube ad scriptwriter specializing in direct response. I need a 30-second non-skippable pre-roll ad for [product/service]. My audience is [ICP]. The ad must: hook in the first three seconds with a bold claim or question, deliver a specific problem the viewer has, present our solution with one concrete outcome, and close with a clear CTA. Write the script as spoken word with pacing notes. Under 80 words total."

Marketing Analytics Prompts

This is one of the most underused prompt categories. Most marketers use ChatGPT for copy. The ones who get 10x more value use it for analysis - turning raw data into decisions.

Campaign Performance Analysis Prompt

"You are a marketing analytics specialist. I ran a campaign with the following results: [paste metrics - impressions, clicks, CTR, conversions, CPA, ROAS, etc.]. Analyze this data and tell me: (1) what's working and why, (2) what's underperforming and the most likely cause, (3) three specific optimizations I should make in the next sprint, and (4) which metrics I should prioritize tracking going forward. Be direct. Give me a prioritized action list, not just an explanation."

The "prioritized action list" instruction is what separates useful analysis from interesting observation. You want to know what to do next, not just what happened.

A/B Test Design Prompt

"You are a conversion rate optimization specialist. I want to A/B test my [landing page / email subject line / ad headline / CTA button]. My current control is [paste current version]. My conversion goal is [goal]. Design three test variants that isolate one variable each. For each variant: explain what you're changing, the hypothesis for why it would improve conversion, what audience or traffic source it's best suited for, and how long to run the test to reach statistical significance given a traffic volume of [X visitors per month]."

Testing without a hypothesis is guessing with extra steps. This prompt forces a hypothesis-first approach, which means you learn something regardless of the outcome.

Customer Feedback Analysis Prompt

"You are a market research analyst. I have collected [X] customer reviews / survey responses / support tickets. Here is the raw data: [paste data]. Analyze it and identify: the top five positive themes customers mention, the top five complaints or friction points, the most common language customers use to describe their problem before finding us, and any patterns that suggest an unmet need or upsell opportunity. Present findings as a bulleted summary with supporting quotes."

This is how you turn qualitative data into messaging. The phrases customers use to describe their own problems? Those are your best subject lines, ad hooks, and landing page headlines. You're not writing copy - you're mining it from the voice-of-customer data that already exists.

Prompts for Specific Marketing Use Cases

Product Launch Prompt

"You are a go-to-market strategist. I am launching [product] in [timeframe]. My ICP is [ICP]. My top three competitors are [A, B, C]. Build a launch plan that covers: a pre-launch buzz campaign (two weeks before), a launch day push, and a post-launch follow-up sequence. For each phase: the goal, the channels, the core message, the key assets needed, and the success metric. Be specific - tell me exactly what to build, not general frameworks."

Pricing Page Copy Prompt

"You are a SaaS conversion copywriter. I need to rewrite my pricing page for [product]. My three tiers are [paste current pricing/features]. My ICP is [ICP]. Rewrite the pricing page copy to: anchor the highest tier as the default choice, highlight the most compelling feature in each tier (not a feature list - one outcome per tier), write a headline for each tier that focuses on who it's for rather than what it costs, and write five FAQ answers that address common pricing objections. No feature bloat. No generic corporate language."

Testimonial and Case Study Prompt

"You are a case study writer. I have the following raw customer testimonial or interview notes: [paste raw content]. Turn this into: (1) a one-sentence social proof quote under 25 words that captures the single best result, (2) a 150-word case study summary in the format of Problem-Solution-Result, (3) three stat-based proof points extracted or inferred from the data provided, and (4) a before/after statement that contrasts where the customer was and where they are now. Format for use on a landing page."

Most testimonials are weak because they capture sentiment instead of outcomes. This prompt forces a result-led structure that makes every testimonial more credible and more persuasive.

Advanced Techniques: Chaining Prompts

The real power of ChatGPT for marketing isn't any single prompt - it's chaining them. Here's how a real workflow looks:

1. Run the ICP prompt - get your three ICPs
2. Feed ICP #1 into the positioning prompt - get differentiated messaging
3. Feed that messaging into the cold email prompt - get tailored outreach copy
4. Feed the best-performing cold email into the follow-up sequence prompt - get a full sequence
5. Run the content pillar prompt for the same ICP - get aligned content strategy
6. Feed each content pillar into the blog post brief prompt - get a full editorial calendar
7. Feed the editorial calendar into the repurposing prompt - get LinkedIn, email, and video content for each post

Each prompt builds on the last. You're not asking ChatGPT to start from scratch every time - you're giving it compounding context. The output gets more specific and useful with every step.

The System Prompt Technique

If you're using ChatGPT's custom instructions feature or building a GPT, you can set a persistent system prompt that carries your brand context across every conversation. Here's what mine includes:

"You are a marketing copywriter and strategist for [brand]. You write in a direct, no-fluff voice. You always lead with the benefit before the feature. You never use the following phrases: [list banned phrases]. You always structure copy for scanability with short paragraphs and clear headers. When I ask for options, give me exactly [X] options and flag the one you'd recommend. My ICP is [ICP]. My offer is [offer]. My differentiator is [differentiator]."

Setting this once means every subsequent prompt in that session starts with full context. You stop explaining your business from scratch every time you open a new chat.

The Iteration Prompt

Most marketers accept ChatGPT's first draft. The ones who get great output iterate. Here's a simple iteration prompt that works for almost any marketing asset:

"That's a good start. Now rewrite version [X] with these specific changes: make the opening more direct, cut the second paragraph entirely, replace the CTA with something lower friction, and make the whole thing 30% shorter. Keep the core argument but tighten every sentence. Do not add new ideas - only refine what's already here."

The instruction to "not add new ideas" is critical. Without it, ChatGPT tends to write a new piece rather than refining the existing one. You'll go in circles. Keep iterations focused on specific changes.

Prompt Mistakes Marketers Make (And How to Fix Them)

I've seen enough bad prompts to know the patterns. Here are the five mistakes I see most often and the fixes:

Mistake 1: No Role Assignment

Writing "write me a marketing plan" instead of "you are a direct-response marketing strategist with experience in B2B SaaS" means ChatGPT draws on the average of everything it knows about marketing - which produces average output. Role assignment shifts it toward expert-level thinking in a specific domain.

Mistake 2: Asking for Too Many Things at Once

"Write me a full campaign including emails, ads, landing page copy, and a 90-day content calendar" is five separate tasks. Break it into five separate prompts. Each one gets more focused output than a single bloated prompt that tries to cover everything.

Mistake 3: No Output Format Specification

If you don't tell ChatGPT what format you want, it will pick one for you. Sometimes that's fine. But if you need a table, ask for a table. If you need a numbered list with specific fields, specify exactly what those fields are. Format instructions cut editing time by half.

Mistake 4: Not Providing Examples

The "few-shot" prompting technique - where you give ChatGPT two or three examples of what you want before asking it to produce the thing - is one of the most powerful prompt engineering techniques available. Instead of describing the tone you want, show it. Paste in a piece of copy that has the right voice and say "write in this style." You'll immediately see the quality jump.

Mistake 5: Taking the First Draft as Final

ChatGPT's first draft is a 7 out of 10 at best for most marketing tasks. The value is in the iteration. Tell it what's wrong, what to keep, and what to change. Run two or three rounds of refinement before you decide whether the output is usable. Marketers who treat the first output as final are leaving a lot of quality on the table.

Tools to Pair With Your ChatGPT Prompts

Prompts are only part of the workflow. Here's the stack I use alongside ChatGPT to turn AI output into executed campaigns:

• Prospect list building: Once your ICP is defined, you need verified contacts. A B2B lead database like this one from ScraperCity lets you filter by job title, industry, company size, and seniority and pull contact data you can feed into outreach tools.
• Email finding: If you have a prospect's name and company but not their email, an email lookup tool fills that gap before your GPT-generated sequence ever goes out.
• Email validation: Before you send a sequence, run your list through ScraperCity's email validator to remove bad addresses. Bounce rates over 3% start damaging your sender reputation, and all the perfect AI-generated copy in the world won't matter if your emails aren't reaching inboxes.
• Outreach sending: Smartlead and Instantly handle sending infrastructure, warm-up, and sequence automation for cold email at scale.
• Personalization at scale: Clay enriches your prospect list with data points you can use in personalized GPT-generated openers.
• CRM and pipeline: Once leads start coming in, Close CRM is built specifically for sales-focused teams that run outbound at volume.

What to Do After You Have the Copy

Prompts generate drafts. Drafts need two things: editing with your real voice and data to validate. Don't publish AI output without reviewing it for accuracy, brand tone, and specificity. Replace any placeholder language, add real examples from your experience, and cut anything that sounds robotic.

The edit pass is where your expertise turns an AI draft into something genuinely useful. ChatGPT doesn't know that your best-performing subject line last quarter was a question. It doesn't know that your ICP hates the word "solution." You do. That knowledge, applied on top of a solid AI draft, is what produces copy that actually converts.

For lead gen specifically, great copy means nothing without a solid prospect list behind it. Once your messaging is tight, make sure your targeting is tight. Use a B2B lead database to pull verified contacts that match your ICP - that's where a lot of campaigns fall apart even when the copy is strong.

Then track everything. Which subject lines get the most replies? Which ad hooks get the lowest CPL? Feed those results back into ChatGPT as context for your next round of prompts. That feedback loop is what separates marketers who get 10x more from AI than everyone else.

If you want hands-on help putting all of this together - building the systems, not just the prompts - that's exactly what I cover inside Galadon Gold.

Building Your Own Prompt Library

One of the highest-leverage investments you can make as a marketer right now is building a personal prompt library. Not a generic list you downloaded from a blog, but a library of prompts you've tested and refined for your specific business, your specific ICP, and your specific voice.

Here's the system I recommend:

1. Start with a template folder. Keep a running document with the prompts from this article as starting templates. Each one has blanks to fill in - [ICP], [offer], [competitor], etc.
2. Track what works. When a prompt produces output you actually use without heavy editing, mark it. Note what made it work. That annotation is worth more than the prompt itself.
3. Version control your context block. Update your company context block every time something meaningful changes - new offer, new ICP insight, new competitor positioning, new banned phrase. A stale context block produces stale output.
4. Build vertical-specific variants. If you serve multiple industries, create a version of your core prompts for each one. The ICP context for a SaaS company is completely different from a professional services firm - and so is the output quality when you specify it.
5. Build a "bad examples" list. Keep a list of the output phrases you keep deleting. "I hope this message finds you well." "Leverage our innovative solution." "At the end of the day." The more specific your banned phrases list, the cleaner your first drafts come out.

The prompts above are a starting point. Build your own library, document what works for your specific audience, and iterate. That's how you turn ChatGPT from a novelty into a real leverage tool for your marketing.

For the full set of my own tested prompts organized by workflow - from market research to lead gen to cold outreach - you've got three free resources to grab: GPT Market Research Prompts, GPT Lead Gen Prompts, and the Cold Email GPT Prompts pack. Each one is built for a specific stage of the pipeline, not a generic dump of AI tips.