The short answer
The best ChatGPT prompts for marketing give OpenAI's ChatGPT a role, your real context (customers, numbers, past work), clear rules and an exact output format. The 12 prompts here cover positioning, customer profiles, campaign plans, content briefs, ad tests, landing pages, competitor research and reporting. They work on every plan, but since 7 October 2026 Plus and higher plans answer with GPT-6 Sol and Free uses the smaller GPT-6 Luna with a 27K-token GPT Instant context window.
Why these ChatGPT marketing prompts work better than the usual lists
Most lists of ChatGPT prompts for marketing are one line long: "write a social post about our product". ChatGPT answers those with the average of everything it has read, which is why the output sounds like every other company. The 12 prompts on this page are longer on purpose. Each one gives ChatGPT a role, the real context (your customer, your numbers, your past work), clear rules and an exact output format. That is the structure OpenAI itself recommends in its prompt engineering guide: describe the identity and goal, give explicit instructions, show examples of the output you want and separate your material with clear markers such as headings or tags like <quotes>.
The prompts are grouped by the job you are doing: strategy and positioning, campaigns and content, ads and landing pages, and research and reporting. Every placeholder in [square brackets] is something only you know. Fill them all in. A prompt with half its placeholders left empty performs like a one-line prompt.
A note on models before you start. Since 7 October 2026, ChatGPT Plus, Pro, Business and Enterprise users get GPT-6 Sol in the Chat tab, while Free and Go users get the smaller GPT-6 Luna. The free plan also has a small GPT Instant context window of 27K tokens, which OpenAI estimates at about 12 pages of input. The research and repurposing prompts below often paste in more than that, so on Free, shorten your source material or split it. Our ChatGPT guide explains every plan in detail.
ChatGPT prompts for marketing strategy and positioning
Strategy prompts fail when you ask ChatGPT for opinions it has no basis for. The three prompts in this group make it work from evidence you supply: customer quotes, your best and worst accounts, and your actual plan and budget. Paste real material, even if it is messy. Interview notes and copied reviews are better input than a tidy summary you wrote yourself, because the customer's words are the raw material for positioning.
The positioning prompt asks ChatGPT to mark any claim it cannot support with [NEEDS PROOF]. Keep that rule in every strategy prompt you write. It turns invented benefits, the most common problem with AI marketing copy, into a visible to-do list. The plan review prompt sets ChatGPT up as a sceptical CMO, because a role with a clear point of view gives sharper feedback than "please review this".
ChatGPT prompts for campaigns and content
Content prompts get better the moment you show ChatGPT what good looks like. Each prompt in this group has a slot for a past piece of content in your voice. Anthropic's prompting guide calls examples "one of the most reliable ways" to steer tone and structure, and the same holds for ChatGPT. One strong example is worth more than five adjectives describing your brand voice.
The repurposing prompt makes ChatGPT cite which part of the source each post came from. That small rule stops it padding posts with generic advice and makes your editing faster, because you can check every claim against the original in seconds. If you run the same campaign prompts every month, save them in a ChatGPT project (available on every plan, including Free) with your brand guide and best examples attached, so you stop pasting the same context again.
ChatGPT prompts for ads and landing pages
Ad copy is where ChatGPT is fastest and where it needs the most constraint. The ad prompt asks for nine variations in a proper test design: three angles, three headlines each, with only one thing changing inside each group. That gives you results you can learn from instead of nine random ads. It also asks for character counts, because ChatGPT is not reliable at counting characters in its head. Always check the counts in the ad platform before you publish.
For landing pages, give ChatGPT one reader and one traffic source. A page written for everyone converts no one, and the model will happily write for everyone unless you tell it otherwise. Paste your analytics too: bounce rate and scroll depth tell ChatGPT where the page loses people, so its rewrite targets the real problem.
ChatGPT prompts for marketing research and reporting
The competitor prompt depends on web search, which shows the sources ChatGPT read. Open the links before you share the table, because a cited page can still be summarised wrongly. For bigger questions, ChatGPT's deep research (limited on Free and Go, expanded on Plus and Pro) reads more pages and writes a longer, cited report. If citations are the whole point, Perplexity puts numbered sources on every answer, including on its free plan.
The reporting prompt uses ChatGPT's data analysis on an uploaded spreadsheet. Data analysis is limited on Free and included from Go upwards. Always ask for the formulas, as the prompt does, and spot-check one number by hand. One more detail: since the GPT-6 update on 7 October 2026, ChatGPT can mix text with visuals and interactive elements, which OpenAI calls Intelligent UI, and Plus and Pro add interactive tables and charts. That is useful for exploring, but if you need a table to paste into a slide, ask for plain text, as this prompt does.
How to adapt these prompts to your brand
- Replace every placeholder with specifics. "Operations leads at 50 to 500 person logistics firms" gives a far better answer than "B2B buyers".
- Add one real example of your best work. A past post, ad or email in your voice does more than any tone description.
- Explain why a rule exists. Anthropic's guide notes that giving the reason behind an instruction helps the model apply it well. "Keep it under 150 words because it is read on phones" works better than "be brief".
- Put long material first and the question last. Both OpenAI and Anthropic recommend separating pasted material with clear markers, and Anthropic advises placing long documents above the instructions.
- Iterate in the same chat. Reply with "Make option 2 more concrete and cut the second paragraph" instead of starting again.
- Save what works. Put your best versions in a project or a skill, or keep them in a shared doc for the team. Our prompt kits bundle tested prompts by role.
Which tool works best for marketing prompts, and why
ChatGPT is the best all-rounder for these prompts if your team already uses it: it handles strategy, copy, spreadsheets and web research in one place, and Plus gives you GPT-6 Sol and projects. Its catch is that the model depends on your plan, so a colleague on Free may get very different answers from the same prompt.
Claude by Anthropic is the stronger pick when the writing itself must be excellent, such as long-form articles, nuanced positioning and brand voice. Claude Opus 5.5 is on paid plans, and the free plan gets Claude Sonnet 5.5. See ChatGPT vs Claude for a full comparison and our Claude prompts for writing.
Gemini suits teams that live in Google Workspace, because Gemini works inside Gmail from the Google AI Plus plan and inside Docs on Google AI Pro. Perplexity is the best choice for the research prompt when every claim needs a visible source.
A privacy reminder: on personal Free, Go, Plus and Pro accounts, ChatGPT can use your chats to train OpenAI's models unless you switch off "Improve the model for everyone" in Settings, Data controls. Do not paste customer data into a personal account. ChatGPT Business and Enterprise do not train on your business data by default. For the full picture of marketing tools, read the best AI for marketing and AI for marketers.
Prompts to copy
Write a positioning statement from real customer words
You are a B2B positioning strategist who has run messaging workshops for software and service companies. Context: We sell [product] to [target customer, e.g. operations leads at 50 to 500 person logistics firms]. Our main competitors are [competitor 1] and [competitor 2]. Below are [number] real quotes from customer interviews, reviews and sales calls. <quotes> [paste quotes] </quotes> Task: 1. Group the quotes into the 3 to 5 problems customers describe most often, using their own words as the group names. 2. For each problem, say which quote proves it best. 3. Write three different positioning statements in the format: For [customer] who [problem], [product] is the [category] that [main benefit], unlike [alternative], which [weakness]. 4. Rank the three by how well the quotes support them and explain the ranking in two sentences each. Rules: use only claims the quotes support. If a statement needs evidence we did not give you, mark it [NEEDS PROOF]. No buzzwords such as innovative, powerful or best-in-class.
You get: Three positioning options ranked by evidence, each tied to the customer quotes that support it
Build an ideal customer profile you can actually target
Act as a demand generation lead. I will give you a list of our best and worst customers from the last 12 months. <best_customers> [paste: company, industry, size, deal value, time to close, why they bought] </best_customers> <worst_customers> [paste: same fields, plus why they churned or stalled] </worst_customers> Compare the two groups and write an ideal customer profile with: firmographics (industry, size, region), the trigger events that make a company ready to buy, the job titles involved and what each one cares about, and three disqualifiers. Then list five places we can find more companies like the best group (LinkedIn filters, events, communities, data sources). Present the profile as a table, then the five sources as a numbered list. Flag any pattern that rests on fewer than three customers.
You get: A one-page ICP table built from your own wins and losses, plus where to find look-alike accounts
Pressure-test a marketing plan before you present it
You are a sceptical CMO reviewing a quarterly marketing plan from a team member. Be direct and specific, not polite. <plan> [paste plan, budget split and targets] </plan> Our goal this quarter is [goal, e.g. 120 sales-qualified leads] with a budget of [budget]. Last quarter we got [result] from [channels]. Review the plan and give me: 1. The three assumptions most likely to be wrong, and what data would prove or disprove each. 2. Any channel where the maths does not add up, with your working. 3. What you would cut and where you would move that money. 4. The five questions my boss is most likely to ask, with a short suggested answer for each. Keep it under 500 words.
You get: The weak spots in your plan, the maths behind them and the questions to prepare for
Turn one campaign idea into a full channel plan
Act as a campaign manager. Our campaign is [campaign name and idea] for [audience]. It runs from [start date] to [end date] and the single goal is [goal]. Channels we use: [e.g. LinkedIn, email newsletter, webinar, paid search]. Our tone is [three adjectives], and here is one piece of past content that sounds right: [paste example]. Create a week-by-week plan as a table with columns: week, channel, asset, message, call to action, owner (leave blank) and the metric that tells us it worked. After the table, write the one-sentence core message every asset must carry and three ways the campaign could fail, with how we would spot each early.
You get: A dated, channel-by-channel campaign table with one core message and early warning signs
Repurpose a long asset into a week of content
Below is our [webinar transcript / report / long blog post]. Turn it into content for one week. <source> [paste] </source> Produce: three LinkedIn posts of 120 to 180 words, each built on a different fact or quote from the source; one newsletter section of 200 words with a subject line; five short social captions under 40 words; and one email to customers inviting them to read the full piece. Every item must use a specific number, example or quote from the source, and you must say which part of the source it came from in brackets. Write for [audience] in a [tone] voice. Do not add statistics that are not in the source.
You get: Ten ready-to-edit pieces of content, each traceable to a specific part of the original
Write a content brief a freelancer can follow
You are a content lead writing a brief for a freelance writer who knows nothing about our company. Topic: [topic]. Target reader: [role and what they already know]. Search phrase we want to rank for: [keyword]. Business goal: [e.g. demo requests from mid-market finance teams]. Our product's relevant feature: [feature]. Pages to link to: [URLs]. Write the brief with: working title and two alternatives, the one question the article must answer in its first 100 words, an outline with H2 and H3 headings, what to include under each heading, three expert or data sources to look for, words to avoid, target length, and a checklist the writer ticks before sending the draft. Keep the brief to one page.
You get: A one-page brief with outline, sources and a pre-submission checklist
Write ad variations that test one idea at a time
You are a performance copywriter. Write ads for [product] aimed at [audience] on [platform, e.g. LinkedIn Ads or Google Search]. What we know: the main pain is [pain]; the proof we can use is [proof, e.g. customer result, award, number of users you can verify]; the offer is [offer]. Write 9 ads in three groups of three. Group A leads with the pain, group B leads with the outcome, group C leads with the proof. Inside each group, change only the headline. Respect the platform's character limits: headline [limit] characters, body [limit] characters, and show the character count after each line. End with a short note on which group you expect to win and what result would prove you wrong.
You get: Nine ads in a clean test design, with character counts and a hypothesis to check
Rewrite a landing page section for one reader
Act as a conversion copywriter. Here is the [hero / pricing / features] section of our landing page and the page's analytics. <section> [paste copy] </section> <data> [e.g. bounce rate, scroll depth, conversion rate, top traffic source] </data> The page is for [one specific reader] who arrives from [traffic source] wanting [goal]. First, list the three things in the current copy that most likely stop that reader, with a reason for each. Then write two rewrites: one that keeps our current structure and one that changes it. Keep headlines under 10 words. Return the result as a before and after table, followed by both rewrites in full.
You get: A diagnosis of what blocks the reader plus two rewrites you can A/B test
Write a product launch announcement in three lengths
We are launching [feature or product] on [date]. What it does: [plain description]. Who it is for: [audience]. The problem it solves: [problem]. Proof: [beta results, quote or number you can verify]. Availability and plans: [details]. Write: a 25-word version for social posts, a 120-word version for the newsletter and a 400-word blog announcement with a headline, a subheading and a clear next step. Use the same core message in all three. Avoid the words excited, thrilled, game-changing and introducing. Write in a [tone] voice that sounds like this example: [paste a past announcement you liked].
You get: A matching set of launch copy in 25, 120 and 400 words
Analyse competitors from their public pages
You are a product marketing analyst. Use web search to read the current homepage, pricing page and two recent blog posts or announcements for each of these competitors: [competitor 1], [competitor 2], [competitor 3]. Today is [date]. For each one, report: who they target, their main claim, the proof they offer, how their pricing is structured (no figures unless shown on their page), and one thing they now say that they did not say six months ago if you can find it. Then give a comparison table and three openings where our product [describe] can stand out. Link every claim to the page you read. If you cannot open a page, say so instead of guessing.
You get: A sourced competitor table and three positioning gaps, with links you can check
Turn campaign numbers into a clear report
I have uploaded our campaign results for [period] as a spreadsheet. Columns: [list columns]. Act as a marketing analyst reporting to a busy leadership team. Calculate cost per lead, conversion rate by channel and the change against [previous period]. Show the formulas you used. Then write a report with: a three-sentence headline summary, the three numbers that matter most, what worked, what did not and why you think so, and two decisions we should make next month. Give the tables as plain text I can paste into a slide, not as an interactive chart.
You get: Checked calculations, a plain text table and a short report with two decisions
Find the real objections in customer feedback
Below are [number] pieces of feedback from [source: sales call notes, support tickets, survey answers, reviews]. <feedback> [paste] </feedback> Classify each item into objections, feature requests, praise and confusion. Count how many items fall into each group. For the five most common objections, quote two real examples each, explain what belief sits behind the objection, and suggest one piece of marketing content that would answer it (for example a comparison page, a case study or a FAQ entry). Present the counts as a table first.
You get: Counted objection themes with real quotes and a content idea to answer each one