The Best ChatGPT Prompts for Marketing: 24 Copy-Paste Prompts for Strategy, Ads, Email, and Analytics

You are a B2B positioning strategist. My product: [ONE-SENTENCE PRODUCT DESCRIPTION]. Customers who love us: [DESCRIBE YOUR 2-3 BEST CUSTOMERS]. What they would use if we did not exist: [COMPETITORS OR STATUS QUO]. Walk through positioning in this order: 1) list our competitive alternatives, 2) identify capabilities we have that the alternatives lack, 3) map each capability to the customer value it enables, 4) define who cares most about that value, 5) name the market category that makes the value obvious fastest. Push back if any of my inputs are vague or unsupported. Finish with one positioning statement in this format: "For [segment] who [need], [product] is the [category] that [key benefit], unlike [alternative] which [limitation]." Then list 3 risks in this positioning I should validate before building messaging on it.

Why it works: It walks the actual positioning methodology step by step instead of jumping to a tagline, and the push-back instruction stops ChatGPT from politely laundering weak inputs into weak positioning.

Act as a fractional CMO for a [INDUSTRY] business with [TEAM SIZE AND WHO DOES MARKETING] and a monthly marketing budget of [BUDGET]. Goal for the next quarter: [SPECIFIC MEASURABLE GOAL, E.G. 200 DEMO REQUESTS]. Current channels and rough results: [WHAT YOU RUN NOW AND WHAT IT PRODUCES]. Build a 90-day plan that picks a maximum of 3 channels, justifying each choice against my budget and team capacity. For each channel give: the core play, weekly time cost, budget allocation, one leading indicator to check at day 30, and explicit kill criteria for dropping it. Do not recommend any channel that requires a dedicated hire. Format as a table plus a one-paragraph narrative I could send to a cofounder.

Why it works: Budget, team size, and kill criteria force realistic channel picks instead of the usual everything-list of SEO, social, email, events, and a podcast.

You are a messaging strategist. Product: [PRODUCT AND WHAT IT DOES]. Primary audience: [AUDIENCE]. Their top 3 pains: [PAINS]. Build a messaging hierarchy: one umbrella value proposition under 12 words, three supporting pillars, and for each pillar exactly two proof points I would need to substantiate it plus the one objection it will trigger. Apply this filter ruthlessly: reject any pillar a competitor could claim word for word, flag it as generic, and propose a sharper alternative that only fits us. Output as a nested outline I can paste into a brand document, with the proof points marked as [HAVE] or [NEED TO GATHER] based on what I told you.

Why it works: The "reject anything a competitor could claim word for word" rule is the filter most messaging docs never pass, and you leave with proof-point homework instead of unsupported claims.

Act as a competitive strategist. I am pasting the homepage copy of my main competitor: [PASTE COMPETITOR HOMEPAGE TEXT]. My product and key strengths: [YOUR PRODUCT AND WHAT YOU DO BETTER]. Analyze their copy for: the audience they are courting, the primary promise, the objections they preempt, and the gaps (what they conspicuously do not say). Then identify 3 differentiation angles I can own that their messaging leaves open. For each angle give: the claim, the evidence I would need to make it credibly, and the risk if they copy it next quarter. Be blunt about where they are stronger than me; I need the honest read, not encouragement. Output as two tables: their teardown, then my angles.

Why it works: It works from real competitor copy you paste rather than ChatGPT's guesses about a company, and the bluntness instruction gets you analysis instead of cheerleading.

You are a customer research analyst. Here are notes on my 5 best customers. For each: [COMPANY OR PERSON, HOW THEY FOUND US, WHAT THEY BOUGHT, WHAT THEY SAY THEY LOVE, ANYTHING ELSE NOTABLE]. Synthesize a working ideal customer profile covering: firmographics or demographics, the trigger event that starts their buying process, the job they hire us for, their decision criteria, and their likely objections. Critical rule: separate what my data clearly supports from what you are inferring, and label every inference explicitly as INFERENCE. End with 5 interview questions I should ask my next customer, targeted specifically at testing your shakiest inferences.

Why it works: Forcing the model to separate evidence from inference stops you from treating a plausible guess as a fact, and the closing questions turn gaps into a research plan.

Act as a jobs-to-be-done researcher preparing me for customer interviews. Product: [PRODUCT]. What I want to learn: [E.G. WHY TRIAL USERS DO NOT CONVERT]. Write a 30-minute interview guide: a 2-line opener that lowers the interviewee's guard, 10 open questions ordered from broad context to specific moments, and 2 follow-up probes per question. Hard rules: no leading questions, no questions answerable with yes or no, no asking people to predict their future behavior, and at least 3 questions that reconstruct a specific past moment ("walk me through the day you decided to look for a solution"). After the guide, list the 3 mistakes I am most likely to make while running it and how to catch myself mid-interview.

Why it works: The no-leading, no-yes/no, no-future-prediction rules encode the hard-won basics of customer research, so the guide holds up with real humans instead of producing polite noise.

You are a voice-of-customer analyst. I am pasting reviews of [YOUR PRODUCT OR A COMPETITOR] from [SOURCE, E.G. G2, AMAZON, REDDIT]: [PASTE 20-50 REVIEWS]. Extract four things: 1) the exact phrases customers use to describe their pain, as verbatim quotes grouped by theme, 2) the moment they describe realizing they needed a solution, 3) what almost stopped them from buying, 4) unexpected use cases. Rank the themes by frequency across the reviews. Do not paraphrase or clean up the quotes; I want customers' actual words to reuse in ads and landing pages. Output as one table per section with columns: verbatim quote, theme, and what I could use it for.

Why it works: Verbatim quotes grouped by frequency-ranked themes are the raw material for copy that sounds like customers instead of marketers, and the no-paraphrasing rule protects that.

Act as a skeptical market analyst. I believe my market is: [DESCRIBE THE MARKET AND WHO BUYS]. My assumptions: [PRICE POINT, ESTIMATED SIZE OF TARGET SEGMENT, EXPECTED PENETRATION]. Build a bottom-up TAM, SAM, and SOM estimate from my inputs, showing every calculation step. Tag each assumption with a confidence level: SOLID, PLAUSIBLE, or HAND-WAVY. Then argue the bear case: three specific reasons this market may be smaller or harder to win than I think. Strict rule: do not invent statistics or cite figures you cannot verify; wherever real data is needed, name the specific source I should check (census data, industry association reports, public filings) instead of fabricating a number.

Why it works: Bottom-up math with labeled confidence levels plus a forced bear case gives you sizing you can defend, and the no-invented-statistics rule keeps the whole thing auditable.

You are a direct response copywriter. Offer: [PRODUCT AND OFFER]. Platform and placement: [E.G. META FEED, LINKEDIN FEED, YOUTUBE PRE-ROLL]. Audience: [AUDIENCE]. Primary pain: [PAIN]. Desired action: [CTA, E.G. START FREE TRIAL]. Generate an ad testing matrix with 4 angles: pain-led, outcome-led, social-proof-led, and objection-led. For each angle write 3 hooks (maximum 12 words each), one body of 40-60 words that fits the placement's conventions, and 2 CTA button variants. Constraints: no exclamation marks, no "unlock" or "supercharge" or "game-changer", reading level around 6th grade, and every factual claim must trace back to something in my inputs; if you need a claim I did not give you, mark it [NEEDS VERIFICATION]. Output as a table with columns: angle, hook, body, CTA.

Why it works: Four angles times three hooks is a real test plan rather than a single ad, and the banned-word list plus claim-traceability rule removes the tells that scream AI copy.

Act as a Google Ads copywriter. Product: [PRODUCT]. Target keyword theme: [KEYWORDS, E.G. "PROJECT MANAGEMENT SOFTWARE FOR CONTRACTORS"]. Searcher intent: [WHAT THEY WANT WHEN THEY SEARCH THIS]. Differentiator: [YOUR EDGE]. Write one responsive search ad: 12 headlines of maximum 30 characters each and 4 descriptions of maximum 90 characters each, showing the character count next to every line. Rules: at least 4 headlines include the keyword theme naturally, at least 2 address price or a common objection, at least 2 contain a specific concrete benefit, no two headlines start with the same word, and no filler like "best solution" or "top rated" unless I gave you proof. Finish by marking the 3 headlines you would pin to position 1 and explaining why.

Why it works: Character counts, keyword coverage quotas, and no-repeat rules mirror how experienced PPC copywriters actually build RSAs, and the pinning recommendation comes with reasoning you can evaluate.

You are a copywriter who works from voice-of-customer data. Here is a real customer quote: "[PASTE THE QUOTE]". Product: [PRODUCT]. Build 3 ads where the quote's core emotion drives the copy: 1) one that opens with the quote verbatim, 2) one that translates its emotion into a question hook, 3) one that dramatizes the before-and-after the quote implies. Each ad: hook, body of about 50 words, CTA. Keep the customer's vocabulary; do not upgrade their plain words into marketing speak. Flag any line where you had to add a fact that is not in the quote or my inputs, so I can verify it before spending money on it.

Why it works: Anchoring on a real quote keeps the emotional core authentic, and the fact-flagging instruction protects you from launching ads built on invented claims.

Act as a paid social strategist. This ad is winning but starting to fatigue: [PASTE THE WINNING AD COPY]. Performance context: [WHAT WORKED: THE HOOK, THE AUDIENCE, THE RESULT]. Generate 6 refresh variants that keep the proven persuasion structure but change the surface: 2 with new openings on the same angle, 2 that flip the format (statement to question, second person to short story, or similar), and 2 that carry the same promise to this adjacent audience segment: [ADJACENT SEGMENT]. For each variant, add a one-line change log stating exactly what you kept and what you changed, so when I test them I can attribute the results to a specific change.

Why it works: It systematically varies the surface while preserving the persuasion structure that earned the original its results, and the per-variant change logs keep your test results attributable.

You are a lifecycle email strategist. Product: [PRODUCT]. New subscribers arrive from: [SOURCE, E.G. LEAD MAGNET, FREE TRIAL SIGNUP]. Sequence goal: [E.G. FIRST PURCHASE, ACTIVATION EVENT]. Write a 5-email welcome sequence. For each email give: the specific job it does in the sequence, send timing, a subject line plus one alternate, preview text, a 120-180 word body, and one CTA. Sequence logic to follow: email 1 delivers the promised value immediately, emails 2 and 3 build the problem and the belief that it is solvable, email 4 handles this objection head-on: [BIGGEST OBJECTION], email 5 makes the direct ask. Voice: [2-3 ADJECTIVES, OR PASTE A WRITING SAMPLE]. No fake urgency and no "just checking in."

Why it works: Assigning every email a defined job produces an argument that builds across five sends instead of five disconnected messages that all ask for the sale.

Act as an email copywriter obsessed with open rates. Email content summary: [WHAT THE EMAIL SAYS AND OFFERS]. Audience and their relationship to us: [E.G. WARM LIST, TRIAL USERS, COLD-ISH NEWSLETTER]. Write 20 subject lines, 4 for each of these 5 mechanisms: curiosity gap, specific benefit, pattern interrupt, personal and casual, and urgency grounded in something real: [THE REAL DEADLINE OR REASON]. Constraints: under 45 characters preferred and show the character count for each, no clickbait the body cannot cash, no all caps, no emoji. Then pick your top 3 and, for each, say which audience segment it will resonate with most and why, so I can pair them correctly in my A/B test.

Why it works: Twenty lines across five named mechanisms turns subject line writing into a structured experiment, and the segment-matching step makes the A/B test smarter than random pairing.

You are a retention marketer. Segment: people who [DEFINE INACTIVITY, E.G. HAVE NOT OPENED IN 90 DAYS, OR CHURNED FROM A PAID PLAN]. What they originally signed up for: [THE ORIGINAL VALUE]. What has changed since they went quiet: [NEW FEATURES, IMPROVEMENTS, OFFERS]. Write 3 win-back emails as one mini-sequence: 1) an honest "here is what is new" with zero guilt-tripping, 2) a concrete re-demonstration of value that shows rather than claims, 3) the clean break: a final email that makes unsubscribing easy and gives one clear reason to stay. Each email: subject line, preview text, body under 150 words. Tone: respectful of their inbox, zero desperation, no "we miss you" theatrics.

Why it works: The three-email arc treats a cold subscriber like an adult, and the clean-break final email is the piece most win-back flows are missing entirely.

Act as a newsletter editor. Audience: [WHO READS AND WHY THEY SUBSCRIBED]. Here is my raw material for this issue: [PASTE BULLET POINTS, LINKS, ROUGH NOTES]. Turn it into a full draft: an opener of 2-3 sentences with an actual point of view (never "welcome back"), the material organized into at most 3 sections with skimmable subheads, one clearly marked promotional block for [WHAT YOU ARE PROMOTING] that takes up no more than 15 percent of the issue, and a one-line sign-off ending in a question that invites replies. Editorial rule: cut anything that serves me more than it serves the reader, and list what you cut and why at the end so I can veto.

Why it works: The 15 percent promo cap and the "cut what serves me more than the reader" rule enforce the editorial discipline that keeps open rates alive issue after issue.

You are a conversion copywriter. Write complete landing page copy for: [PRODUCT OR OFFER]. Visitors arrive from: [TRAFFIC SOURCE AND THE AD OR LINK MESSAGE THEY CLICKED]. The one action I want: [CONVERSION GOAL]. Structure: hero (headline under 10 words, subhead, CTA), a problem section written in the visitor's own words, a solution section with 3 benefit blocks (each: benefit headline, 2-sentence support, the feature that proves it), a social proof section where you specify exactly what proof I need to supply rather than inventing testimonials, an objection section as 3 short Q&As, and a final CTA with this risk reversal: [GUARANTEE, FREE TIER, OR TRIAL]. Non-negotiable rule: the hero headline must match the message visitors clicked, so the promise carries through. Label every section for my page builder.

Why it works: Message match between the ad and the hero is the highest-leverage conversion rule there is, and here it is baked into the structure instead of left to chance. The proof placeholders also stop ChatGPT from fabricating testimonials.

Act as a CRO specialist. Current landing page headline: "[CURRENT HEADLINE]". What the page offers: [OFFER]. Who lands here and from where: [AUDIENCE AND TRAFFIC SOURCE]. Their awareness stage: [UNAWARE / PROBLEM-AWARE / SOLUTION-AWARE / PRODUCT-AWARE]. Write 12 headline alternatives grouped by strategy: 3 outcome-specific, 3 problem-agitation, 3 objection-preempting, 3 curiosity-with-substance. Each under 10 words, each with a matching subhead. Then critique my current headline honestly against your strongest alternative: what it does well, where it leaks attention, and whether it even fits my traffic's awareness stage. Ban superlative soup: no "ultimate," "best," or "revolutionary."

Why it works: Awareness stage changes which headline strategies are even viable, and grouping variants by strategy means your test results teach you something instead of just crowning a random winner.

You are a SaaS pricing page specialist. Plans: [LIST PLANS, PRICES, AND KEY DIFFERENCES]. Target buyer: [WHO PAYS AND WHO USES IT]. The plan I want most people to choose: [PLAN]. Write: a pricing page headline that frames value rather than cost, a one-line description per plan that helps the right buyer self-select, a "which plan is right for me" section with 3 short scenarios mapped to plans, and an FAQ handling these 5 objections: [E.G. PRICE TOO HIGH, CONTRACT FEARS, SWITCHING COST, "CAN I JUST USE FREE TOOLS", SECURITY]. Constraints: no "unbeatable value" language, and name each plan's honest trade-off; buyers trust pricing pages that admit limits, so tell them who each plan is NOT for.

Why it works: Self-selecting plan descriptions and honest trade-offs handle the hesitations that actually stall upgrades, which is more valuable than another feature comparison grid.

Act as a conversion copy auditor. Here is the page: [PASTE FULL PAGE COPY]. Context: the page goal is [GOAL], the audience is [AUDIENCE], and traffic comes from [SOURCE]. Audit in this exact order: 1) clarity: could a stranger say what we sell within 5 seconds, 2) specificity: flag every vague claim and rewrite the worst one as a concrete example, 3) flow: identify exactly where attention dies, 4) proof: list every claim that lacks evidence, 5) CTA logic: does the ask match how ready this visitor actually is. Score each lens 1-5 and give the single highest-impact fix per lens. Then rewrite the weakest section in full. Be direct; I would rather hear it from you than from the bounce rate.

Why it works: The lens ordering mirrors how professional copy audits run (clarity problems make every other fix pointless), and the mandatory rewrite of the weakest section makes it actionable, not just critical.

You are a marketing analyst who explains data plainly. Here are my numbers for [CHANNEL OR CAMPAIGN] over [PERIOD]: [PASTE METRICS: SPEND, IMPRESSIONS, CLICKS, CONVERSIONS, REVENUE. INCLUDE THE PRIOR PERIOD IF YOU HAVE IT]. Business context: [WHAT CHANGED: NEW CREATIVE, SEASONALITY, PRICE CHANGE, SITE CHANGES]. Tell me: 1) what actually changed versus the prior period and by how much, 2) the 2-3 most plausible explanations, ranked by likelihood given my context, 3) for each explanation, the specific data that would confirm or kill it, 4) the one action you would take this week and what result would validate it. Do not restate my numbers back at me; interpret them. Explicitly flag any conclusion this data is too thin to support.

Why it works: Ranked explanations with confirm-or-kill data requests turn a metrics dump into a decision, and the thin-data flag keeps ChatGPT from manufacturing confidence your sample size does not justify.

Act as a marketing lead writing a monthly report for [AUDIENCE: CEO, CLIENT, OR BOARD] who cares most about [THEIR PRIORITY, E.G. PIPELINE, ROI, GROWTH EFFICIENCY]. Raw results: [PASTE KEY METRICS AND WHAT SHIPPED THIS MONTH]. Write a one-page narrative: a 3-sentence executive summary that leads with the single most important development (good or bad), a "what worked" section that ties results to specific decisions we made, a "what did not work and why" section that owns the misses without spin, and a "next month" section with 3 commitments and the metric that defines success for each. Under 400 words total. No jargon my [AUDIENCE] would have to look up. Use my numbers exactly as given; do not extrapolate or round favorably.

Why it works: Leading with the most important development and owning misses without spin is what separates reports that build trust from reports that get skimmed and forgotten.

You are a growth experimenter. I want to test: [WHAT YOU WANT TO CHANGE AND WHY YOU THINK IT WILL WORK]. Current baseline: [METRIC AND VALUE, E.G. LANDING PAGE CONVERTS AT 2.1 PERCENT ON ROUGHLY 4,000 VISITS PER MONTH]. Design the test: a hypothesis in the format "we believe X because Y; we will know we are right if Z," a primary metric and one guardrail metric, a realistic estimate of how long I need to run it at my traffic level to trust the result (show your reasoning and state your assumptions plainly instead of faking statistical precision), and the 3 most likely ways this test could mislead me, such as novelty effects, seasonality, or segment mix. Then write the decision rules before we start: what result ships it, what kills it, and what means run it again.

Why it works: Writing decision rules before the test runs is the discipline that prevents post-hoc rationalization, and the misleading-results list saves you from shipping a false winner.

Act as a marketing ops specialist. My team runs campaigns across [CHANNELS, E.G. GOOGLE ADS, META, EMAIL, PARTNERSHIPS]. Our current tagging is inconsistent: [DESCRIBE THE MESS, E.G. MIXED CASING, NO NAMING CONVENTION, DUPLICATE SOURCES]. Design a UTM and campaign naming taxonomy the team will actually stick to: conventions for source, medium, campaign, content, and term; a naming formula for campaigns using [AN EXAMPLE CAMPAIGN] as the worked example; casing and separator rules; 10 fully filled-in example URLs across my channels; and a short "never do this" list of the mistakes that silently corrupt reporting. Output as a reference document I can paste into our team wiki as-is.

Why it works: A convention with worked examples and a "never do this" list is the difference between a taxonomy that gets followed and a doc that gets ignored. Clean tagging is the prerequisite for every report you will ever run.