Most marketing prompts fail before ChatGPT answers. They ask for a campaign, article or ad without supplying the customer research, product facts, brand examples or performance data a competent marketer would need.
The 21 prompts below start with those inputs. Each one produces a specific marketing artifact you can review: a positioning brief, campaign plan, content roadmap, email sequence, ad test matrix, landing-page audit or performance report.
Replace every bracketed field, attach the relevant source material and keep human approval between the draft and the customer. ChatGPT is quick. It is also perfectly willing to make unsupported marketing claims with excellent grammar.
The best ChatGPT marketing prompts at a glance
Use this shortcut if you already know where the work is stuck:
- Need clearer customers and positioning: prompts 1 to 4
- Need a plan or campaign direction: prompts 5 to 7
- Need content, SEO or social assets: prompts 8 to 13
- Need email, ads or landing-page improvements: prompts 14 to 18
- Need better measurement and decisions: prompts 19 to 21
| # | Marketing task | What ChatGPT produces |
|---|---|---|
| 1 | Mine customer language | Evidence-backed themes and quotations |
| 2 | Define priority customer segments | Segment scorecard and ICP |
| 3 | Analyze competitor positioning | Positioning comparison and gaps |
| 4 | Build a message hierarchy | Approved claims and proof map |
| 5 | Create a 90-day marketing plan | Prioritized execution roadmap |
| 6 | Write a campaign brief | Review-ready campaign brief |
| 7 | Plan a product launch | Phased launch plan and checkpoints |
| 8 | Build an SEO content roadmap | Topic clusters and page priorities |
| 9 | Create an SEO content brief | Writer-ready article brief |
| 10 | Draft a marketing article | Source-grounded article draft |
| 11 | Repurpose approved content | Channel-specific content package |
| 12 | Document a brand voice | Practical writing guide and examples |
| 13 | Build a social content calendar | Four-week publishing plan |
| 14 | Create a nurture sequence | Lifecycle email sequence |
| 15 | Draft an email campaign | Campaign copy and QA checklist |
| 16 | Write Google Search ad assets | Responsive search ad asset set |
| 17 | Design paid social experiments | Test matrix with hypotheses |
| 18 | Audit a landing page | Message-match and conversion fixes |
| 19 | Diagnose funnel performance | Bottleneck analysis and next actions |
| 20 | Write a campaign report | Leadership-ready performance readout |
| 21 | Design an A/B test | Experiment plan and decision rule |
How to get better marketing output from ChatGPT
OpenAI’s current prompt engineering guidance recommends clear, specific instructions and iterative refinement. For marketing work, the quality of the source material matters just as much as the wording of the prompt.
Provide the information that could change the answer:
- The offer, audience and commercial goal
- Customer research and exact customer language
- Approved product facts, proof and claims
- Brand voice examples and channel requirements
- Current performance data and measurement definitions
- Budget, capacity, deadlines and approval rules
- What ChatGPT must leave unknown instead of inventing
For recurring work, save this context block and paste it before the relevant prompt:
MARKETING CONTEXT
Company: [name and one-sentence description]
Offer: [product or service]
Priority audience: [customer segment and buying situation]
Positioning: [category, promise and differentiator]
Approved proof: [results, customers, research, credentials or none]
Brand voice: [description plus links or pasted examples]
Commercial goal: [pipeline, sales, retention, launch or another outcome]
Active channels: [channels]
Markets and languages: [locations and languages]
Constraints: [budget, capacity, regulation, deadlines or platform limits]
Claims to avoid: [unsupported, prohibited or unapproved claims]
Sources of truth: [documents, dashboards, CRM fields or links]
When prompts are not enough
If the task depends on many files or frequently changing campaign data, stop copying fragments into chat. Give your AI assistant controlled access to approved sources such as your CRM, analytics exports, campaign reports, drive or data warehouse.
OpenAI describes ChatGPT Work for marketing as a way to combine customer insights, campaign context and brand standards. Where available and enabled, ChatGPT apps available through the Plugin Directory can connect approved tools and data to marketing workflows. Other capable AI assistants offer similar file and tool integrations.
For multi-source market or competitor research, deep research can produce a cited report from selected sources. For campaign or reporting data, upload or connect the actual export. ChatGPT’s data analysis environment does not fetch external web data by itself.
Availability depends on the plan, workspace, role, region and administrator settings. Grant only the access required for the task. Keep publishing, budget changes, customer communication and other consequential actions behind human approval.
Customer research and positioning prompts
1. Mine customer language from research
Use this when you have interviews, reviews, sales calls, surveys or support conversations and need messaging grounded in what customers actually say.
Analyze the customer research below for marketing insights.
Offer: [product or service]
Priority audience: [segment]
Research source: [paste transcripts, reviews, survey responses, sales notes or support conversations]
Research period: [dates]
Marketing decision: [homepage, campaign, positioning, content or another decision]
Use only the supplied material. Do not merge quotations, correct grammar inside quotes or invent customer statements. Remove personal information from the output.
Return:
1. Repeated problems, desired outcomes, triggers, objections and alternatives
2. A frequency count for each theme
3. Exact quotations supporting the strongest themes, labeled by source
4. Differences between customer segments or buying situations
5. Claims the evidence supports and claims that need more research
6. 10 messaging angles written in the customer's language without copying full quotations
7. 5 follow-up research questions
Separate explicit customer statements from your interpretation. Note sample-size limits and obvious research bias.
Check before use: Confirm every quotation against the original source and remove anything that could identify a customer without permission.
2. Define priority customer segments from evidence
This prompt ranks customer segments using commercial evidence rather than inventing a persona with suspiciously detailed weekend habits.
Identify and prioritize the customer segments this marketing team should focus on.
Offer and pricing: [details]
Current customers: [CRM export or summary]
Won and lost deal evidence: [notes or data]
Customer research: [interviews, reviews or surveys]
Channel performance: [qualified leads, conversion, revenue and retention by segment]
Business constraints: [capacity, geography, margin or sales model]
Current segment hypothesis: [what we believe today]
Use only supplied evidence. Do not invent demographics, budgets, market size or buying behavior.
Return:
1. Distinct customer segments visible in the evidence
2. A scorecard for problem severity, fit, willingness to act, reachability, commercial value and evidence confidence
3. The strongest primary segment and any credible secondary segment
4. Jobs, triggers, objections and alternatives for each priority segment
5. Qualification and disqualification criteria
6. A concise ICP statement for marketing and sales
7. Missing evidence and a research plan to close the gaps
Show which source supports each conclusion. If the data cannot support a priority, say so.
Check before use: Review segment economics with sales or finance before reallocating budget.
3. Analyze competitor positioning
Use this to compare what competitors claim, which customer they address and where the category has become interchangeable.
Analyze competitor positioning for this market.
Our company: [offer, audience and current positioning]
Competitors: [names and URLs]
Target market: [segment and geography]
Approved evidence: [homepage copy, pricing, reviews, sales feedback and product details]
Research date: [date]
If you have web access, use current first-party competitor pages and cite each factual claim with its URL. If not, use only the material I provide. Do not infer revenue, market share, customer count or product capabilities.
Return:
1. A table comparing category, target customer, core promise, proof, pricing position, CTA and notable limitations
2. Claims and phrases repeated across the market
3. Customer needs each competitor appears to prioritize
4. Credible positioning gaps worth researching
5. Three positioning directions for us
6. The audience, promise, proof required and tradeoffs for each direction
7. A recommendation and the evidence that would change it
Distinguish competitor claims from independently verified facts.
Check before use: Validate the proposed gap with customers. A blank space on a positioning map may simply be a place nobody wants to stand.
4. Build a message hierarchy and proof map
This creates a shared hierarchy for a website, campaign or sales deck while blocking unsupported claims.
Build a message hierarchy for this offer.
Offer: [description]
Priority audience and buying situation: [details]
Customer evidence: [quotes, interviews or sales notes]
Product facts: [approved capabilities]
Proof: [case studies, results, credentials or none]
Competitive context: [approved findings]
Desired action: [purchase, demo, trial, enquiry or another CTA]
Brand voice: [guidance and examples]
Claims we cannot make: [list]
Return:
1. The audience problem in plain language
2. A one-sentence value proposition
3. The main promise and three supporting messages
4. A proof map showing evidence for every message
5. Objections and honest responses
6. Homepage or campaign headline options using different angles
7. CTA hierarchy
8. Claims that remain unsupported
Do not turn features into benefits unless the supplied evidence supports the connection. Leave missing proof as [PROOF NEEDED].
Check before use: Have product, sales and legal owners approve the claims before they become campaign copy.
Related resource: GPTBot’s unique selling proposition prompt.
Marketing strategy and campaign planning prompts
5. Create a focused 90-day marketing plan
Use this to connect one commercial objective to a realistic plan with owners, capacity and measurement.
Create a 90-day marketing plan for this business.
Offer: [description]
Priority audience: [segment]
Commercial objective: [specific outcome]
Baseline: [traffic, qualified leads, conversion, revenue, retention and active channels]
Budget: [amount]
Team capacity: [people and available hours]
Existing assets: [content, list, proof, partnerships and tools]
Constraints: [market, brand, regulation, seasonality or deadline]
Prioritize the few activities most likely to affect the commercial objective. Do not include a channel merely because it is popular.
Return:
1. One measurable objective and supporting indicators
2. Audience, buying situation and core message
3. Ranked channel choices with rationale
4. A week-by-week execution calendar
5. Owners, dependencies, budget and capacity by activity
6. Experiments with hypotheses, thresholds and stop rules
7. A weekly measurement dashboard
8. Major risks and missing inputs
9. What not to do during these 90 days
Do not invent benchmarks, CPCs, conversion rates or expected returns.
Check before use: Make sure the plan fits actual production and sales capacity. More leads are not useful when nobody can follow up.
Related resource: GPTBot’s content marketing strategy prompt.
6. Write a campaign brief
This turns approved research and strategy into one document that creative, media and stakeholders can use.
Create a campaign brief from the information below.
Campaign objective: [outcome]
Offer: [product, service or event]
Priority audience: [segment and buying situation]
Customer insight: [supported insight]
Core message and proof: [approved messaging]
Channels: [approved channels]
Budget: [amount]
Timing: [dates]
Required assets: [known deliverables]
Brand and legal constraints: [rules]
Measurement: [primary conversion and supporting metrics]
Approvers: [roles]
If essential information is missing, ask up to 5 questions in one batch before drafting.
Return:
1. Background and business objective
2. Audience and insight
3. Single-minded proposition
4. Reasons to believe
5. Desired customer action
6. Channel roles and asset requirements
7. Mandatories and exclusions
8. Timeline, owners and approval gates
9. Measurement plan
10. Open questions and risks
Keep the brief concise enough to use. Do not disguise undecided strategy as polished copy.
Check before use: Confirm one accountable owner and one primary conversion before production starts.
7. Plan a product or service launch
This creates a phased launch plan with audience, assets, readiness gates and contingency decisions.
Create a launch plan for this offer.
Offer: [description]
Launch type: [new product, feature, service, market or relaunch]
Priority audience: [segment]
Customer evidence: [research and demand signals]
Positioning and proof: [approved messages]
Launch date: [date]
Markets and channels: [list]
Budget and team: [resources]
Product, sales and support readiness: [status]
Dependencies: [list]
Known risks: [list]
Return:
1. Launch objective and target audience
2. Readiness checklist with owners
3. Pre-launch, launch and post-launch phases
4. Channel and asset plan
5. Internal enablement for sales, support and partners
6. Measurement plan with baseline, primary metric and guardrails
7. Decision gates and fallback options
8. Risk register
9. The first 10 actions in order
Do not invent availability, features, customer demand, pricing or launch commitments.
Check before use: Delay the campaign if product, support or tracking readiness cannot support the promise.
Content, SEO and social media prompts
8. Build an SEO content roadmap
This maps search demand to business relevance and existing content instead of producing a large list of vaguely related keywords.
Create an SEO content roadmap for this business.
Offer and commercial priorities: [details]
Audience: [segments]
Markets and languages: [locations]
Website and existing content: [URLs or export]
Search performance: [Search Console data]
Keyword data: [approved tool export]
Competitors: [URLs]
Publishing capacity: [articles or updates per month]
Available expertise and proof: [people, data and examples]
Use supplied tools for search volume and difficulty. Do not invent metrics. Group queries by shared search intent, not similar wording.
Return:
1. Existing content gaps, overlap and cannibalization risks
2. Three to five topic clusters tied to offers and customer problems
3. Pillar and supporting pages for each cluster
4. Primary query, intent, funnel stage and internal links for every page
5. Pages to create, update, consolidate or leave alone
6. A prioritized 12-week roadmap
7. Evidence or first-hand experience each page needs
8. Conversion action for each page
Rank opportunities by business relevance, evidence, realistic competitiveness and ability to create a better answer.
Check before use: Google says generative AI can help with research and structure, but scaled pages without added value may violate its spam policies. Review Google’s guidance on generative AI content before turning a roadmap into a content factory.
Related resource: GPTBot’s SEO content roadmap prompt.
9. Create a writer-ready SEO content brief
Use this after choosing a target query. The prompt defines what the page must answer, prove and connect to.
Create an SEO content brief for this page.
Target query: [query]
Audience and search situation: [who is searching and why]
Search intent: [informational, commercial, transactional or mixed]
Offer and desired conversion: [details]
Current ranking pages: [URLs or extracted headings]
First-party expertise: [experience, data, examples and opinions]
Existing internal pages: [URLs]
Brand and legal constraints: [rules]
Do not copy competitor structures blindly. Identify what a useful answer requires and where current results are weak.
Return:
1. Recommended title, H1 and meta description
2. Search intent and reader questions
3. Clear angle and promise
4. H2 and H3 outline with purpose for each section
5. Evidence, examples and first-hand input required
6. Claims that need verification
7. Internal-link opportunities
8. Visual or table opportunities
9. Conversion path and CTA
10. Editorial QA checklist
Avoid sections that exist only because every competitor has them. Do not invent facts, statistics or expert experience.
Check before use: The brief should create a better page, not a longer imitation.
Related resource: GPTBot’s SEO blog post outline generator.
10. Draft a source-grounded marketing article
Use this when the topic, audience and evidence are approved and you need a complete first draft rather than another empty outline.
Draft a source-grounded marketing article.
Topic and working thesis: [details]
Audience and reader problem: [details]
Business objective and CTA: [goal]
Target length: [range]
Source pack: [paste notes, links, reports and approved research]
First-party expertise: [SME notes, experience, data and examples]
Approved customer stories or quotations: [sources and permissions]
Brand voice: [guidance and examples]
Claims or topics to avoid: [list]
Internal links: [approved URLs]
Use only the supplied sources for factual claims. If the evidence is insufficient, ask up to 5 questions in one batch before drafting.
Return:
1. A clear thesis and reader promise
2. A concise outline
3. The complete article with sentence-case headings and short paragraphs
4. Source markers beside every externally verifiable claim
5. First-party examples and a credible counterargument where relevant
6. A fact-check queue for unresolved claims
7. Suggested title, meta description and excerpt
8. Internal-link and visual opportunities
Do not invent statistics, quotations, case studies, credentials, customer outcomes or citations. Mark missing evidence as [SOURCE NEEDED].
Check before use: Verify every claim and quotation, add real editorial judgment, and remove any paragraph that merely sounds complete. Google rewards useful, original content, not the speed of the drafting tool.
Related resource: GPTBot’s SEO blog post expert prompt.
11. Repurpose approved content across channels
Use this to adapt one approved source without flattening every channel into the same caption.
Repurpose this approved source into channel-specific marketing assets.
Source content: [paste final article, webinar, report, video transcript or case study]
Audience: [segment]
Objective: [educate, capture demand, nurture or another goal]
Channels: [email, LinkedIn, X, Instagram, YouTube or others]
Brand voice: [guidance and examples]
Approved claims and CTA: [details]
Channel constraints: [format and length]
Preserve the source's factual meaning. Do not add statistics, examples, customer results or claims that are absent from the source.
Return for each channel:
1. The channel-specific audience behavior and angle
2. Recommended format
3. Final draft
4. CTA
5. Suggested visual or clip
6. Claims retained from the source
Vary the hook, structure, pacing and CTA by channel. Do not merely shorten the same paragraph several times.
Check before use: Compare every factual statement with the approved source before publishing.
Related resource: GPTBot’s content paraphraser prompt.
12. Extract a practical brand voice guide
This turns approved examples into observable writing rules instead of a list of adjectives that could describe a coffee shop or a bank.
Create a practical brand voice guide from these examples.
Business and audience: [context]
Channels: [where the voice appears]
Approved samples: [paste 3 to 10 examples]
Rejected samples: [paste examples and explain why]
Required terminology: [terms]
Formatting rules: [known rules]
Analyze observable patterns only.
Return:
1. A concise voice summary
2. Sentence length, vocabulary, rhythm and paragraph patterns
3. How the voice changes by channel
4. Words and constructions to use
5. Cliches, habits and phrases to avoid
6. Punctuation and formatting rules
7. Five before-and-after rewrites
8. A pre-publication checklist
Quote short samples to support each rule. Preserve factual meaning in every rewrite. Do not assign a personality trait unless the evidence supports it.
Check before use: Keep the approved samples with the guide. Examples teach the model more than “confident, human and approachable.”
Related resource: GPTBot’s writing style analyzer prompt.
13. Build a four-week social content calendar
This creates a manageable calendar from real marketing priorities and reusable source material.
Create a four-week social content calendar.
Business and offer: [context]
Priority audience: [segment]
Commercial goal: [goal]
Platforms: [list]
Approved source material: [articles, research, customer questions, launches and proof]
Brand voice: [guidance]
Publishing capacity: [posts and production hours per week]
Available formats: [text, image, carousel, short video or others]
CTA options: [approved CTAs]
Constraints: [claims, approvals or events]
Return:
1. Weekly themes tied to the commercial goal
2. A calendar with date, platform, format, audience need, hook, core point, source, CTA, owner and status
3. A balanced mix of education, proof, product, conversation and conversion
4. Production dependencies
5. Repurposing opportunities
6. Success metrics by post type
Do not invent customer stories, trends or cultural moments. Avoid repeating one idea with cosmetic wording changes.
Check before use: Remove any post that has no source, audience need or reason to exist.
Related resource: GPTBot’s social media schedule prompt.
Email, paid ads and landing-page prompts
14. Create a lifecycle nurture sequence
Use this to move a specific audience from one known stage to the next without sending five versions of “just checking in.”
Create a lifecycle email sequence.
Audience and entry trigger: [who enters and why]
Current awareness stage: [stage]
Desired next stage or action: [goal]
Offer: [description]
Known objections and questions: [research]
Approved proof and content: [sources]
Sequence length and timing: [requirements]
Brand voice: [guidance]
Consent and suppression rules: [rules]
Return:
1. Sequence strategy and purpose of each email
2. Timing and exit conditions
3. Subject line and preview text options
4. Draft body for each email
5. One CTA per email
6. Personalization fields and fallback values
7. Required proof, links and tracking
8. QA checklist
Use only approved claims and sources. Do not invent scarcity, customer results, deadlines or personal details.
Check before use: Confirm consent, sender identity, opt-out handling and local requirements. The FTC’s CAN-SPAM compliance guide is a useful US reference, not a substitute for the laws in your market.
Related resource: GPTBot’s email campaign strategist prompt.
15. Draft a complete email campaign
This produces a review package for a single campaign, including the brief, variants and QA.
Draft an email campaign from the approved brief.
Campaign objective: [goal]
Audience and segment rule: [details]
Offer or message: [details]
Customer insight: [evidence]
Approved proof: [claims and sources]
Primary CTA: [action]
Send timing: [date and timezone]
Brand voice: [guidance and examples]
Required legal or footer content: [requirements]
Past performance: [relevant results]
Return:
1. A one-paragraph campaign strategy
2. Three subject lines using distinct angles
3. Preview text for each
4. The complete email body
5. CTA copy
6. A plain-text version
7. Two meaningful test ideas
8. A pre-send QA checklist
Keep the message focused on one audience and one action. Do not invent urgency, discounts, testimonials or product capabilities.
Check before use: Test links, personalization fallbacks, mobile rendering, tracking and suppression before sending.
Related resource: GPTBot’s email marketing campaign prompt.
16. Write Google responsive search ad assets
Use this after the keyword theme, landing page and conversion measurement are defined.
Create responsive search ad assets for this Google Ads ad group.
Target market and language: [details]
Keyword theme and match types: [list]
Search intent: [buying situation]
Landing page: [URL or approved copy]
Offer: [description]
Verified differentiators: [proof]
Required terms or disclaimers: [text]
CTA: [action]
Claims to avoid: [list]
Return:
1. Fifteen distinct headlines of no more than 30 characters
2. Four distinct descriptions of no more than 90 characters
3. Two path-field options of no more than 15 characters each
4. A table showing exact character count, message angle and keyword relevance for every asset
5. Recommended pinning only where a term or disclaimer must always appear
6. Claims requiring approval
Use only verified landing-page facts. Do not invent prices, savings, response times, certifications or customer results. Check every character count mechanically before returning the table.
Check before use: Confirm the assets against Google’s current responsive search ad specifications and the live landing page. Platform limits are not the place for creative interpretation.
17. Design a paid social experiment matrix
This creates testable creative directions rather than a pile of ad variants with no learning agenda.
Design a paid social experiment matrix for this offer.
Platform: [Meta, LinkedIn, TikTok or another platform]
Objective and conversion event: [goal]
Audience: [approved targeting and exclusions]
Offer: [description]
Customer research: [problems, outcomes, triggers and objections]
Approved proof and assets: [sources]
Budget and test period: [amount and dates]
Baseline performance: [known data]
Brand and policy constraints: [rules]
Return:
1. The core audience hypothesis
2. Four distinct message angles tied to research
3. Creative concept, format, hook, body and CTA for each angle
4. One primary variable per test
5. Test cells, budget logic and minimum runtime assumptions
6. Primary metric, guardrails and decision rule
7. Learning agenda and next test for each possible result
8. Claims and platform requirements to verify
Do not invent benchmark CPMs, conversion rates, audience size or expected lift.
Check before use: Confirm current platform specifications, policies and event tracking before launch.
18. Audit landing-page message match and conversion clarity
Use this when traffic arrives but the page does not clearly continue the promise made by the ad, email or search result.
Audit this landing page for message match and conversion clarity.
Traffic source and message: [ad, email, query or referral copy]
Landing page: [URL or paste rendered copy]
Priority audience: [segment]
Offer: [description]
Primary conversion: [action]
Available proof: [approved evidence]
Current performance: [sessions, conversion and segment data]
Constraints: [brand, legal, layout and engineering]
Evaluate:
- Message match with the traffic source
- Clarity of audience, problem, offer and outcome
- Proof and objection handling
- CTA and form friction
- Unsupported claims
- Mobile scanning and heading structure
Return:
1. The five highest-impact problems, ranked
2. Evidence from the current page for each problem
3. Revised message hierarchy
4. New copy only for weak sections
5. Form and CTA recommendations
6. Three testable hypotheses with one variable each
7. Proof or information required before publishing
Do not invent testimonials, urgency, results, customers or capabilities. Preserve strong sections.
Check before use: Verify that the conversion fires only after a successful action, not merely when someone clicks the button.
Related resource: GPTBot’s website copy optimization prompt.
Analytics and optimization prompts
19. Diagnose a marketing funnel
This prompt identifies where the funnel is changing and what evidence is still missing before someone declares the copy responsible.
Diagnose this marketing funnel.
Business model and offer: [context]
Funnel stages and definitions: [stages]
Reporting period and comparison period: [dates]
Segmented data: [channel, campaign, audience, device, geography and cohort]
Spend and revenue data: [figures]
Tracking changes or known data issues: [notes]
Recent campaign, pricing, product or sales changes: [context]
Preserve all supplied figures. Show calculations and reconcile totals.
Return:
1. Funnel table with volume and conversion between each stage
2. Material changes by segment and period
3. The first meaningful bottleneck
4. Facts, hypotheses and unknowns in separate sections
5. Possible causes ranked by evidence strength
6. Additional data needed
7. Recommended actions and experiments
8. Risks of optimizing the wrong stage
Do not claim causation from correlation. Do not invent benchmarks or fill missing values.
Check before use: Have the analytics owner confirm event definitions and data quality before acting on the diagnosis.
Related resource: GPTBot’s marketing funnel framework prompt.
20. Turn campaign data into a performance report
Use this to create a concise readout that separates outcomes, activity and unresolved questions.
Create a marketing campaign performance report.
Campaign objective and brief: [paste]
Reporting period: [dates]
Baseline or comparison: [figures]
Channel data: [spend, reach, qualified actions, pipeline, revenue and other approved metrics]
Audience and creative breakdowns: [data]
Attribution model and known limitations: [details]
Business context: [launches, tracking changes, seasonality or sales capacity]
Audience for the report: [leadership, client or team]
Return:
1. Executive summary in no more than five bullets
2. Results against objective and baseline
3. Channel, audience and creative breakdowns
4. What changed and what cannot be concluded
5. Winners and underperformers using the agreed metrics
6. Data-quality and attribution caveats
7. Recommendations ranked by expected learning or business impact
8. Decisions required from the audience
Use supplied data only. Show calculations and avoid presenting attributed revenue as proven incrementality.
Check before use: Make sure each recommendation follows from the data rather than from the channel team’s preferred explanation.
21. Design an A/B test with a decision rule
This creates a proper experiment around one decision instead of launching variants and choosing whichever dashboard number looks friendliest.
Design an A/B test for this marketing decision.
Decision to make: [decision]
Current experience and baseline: [details and data]
Proposed change: [variant]
Audience and traffic source: [segment]
Primary metric: [metric and definition]
Guardrail metrics: [metrics]
Expected practical effect: [minimum change worth acting on]
Available traffic or sample: [amount]
Test duration constraints: [dates]
Known sources of bias: [seasonality, campaigns, device mix or others]
Return:
1. Hypothesis and rationale
2. Control and variant definition
3. One primary variable being tested
4. Eligibility, assignment and exclusion rules
5. Primary and guardrail metrics
6. Required sample-size inputs and what remains unknown
7. Minimum runtime and contamination risks
8. Decision rule for win, loss and inconclusive result
9. QA checklist before launch
10. Follow-up analysis after the test
Do not invent statistical significance, sample size or expected lift. If inputs are insufficient, list what an analyst must calculate.
Check before use: Agree on the decision rule before viewing results. Otherwise every test becomes an elaborate permission slip for what someone already wanted.
Review marketing output before it goes live
Before publishing, sending or spending money, check:
- Every product fact, quotation, result and comparison has a source
- Customer data and connected tools were used according to policy
- Claims, testimonials and endorsements are truthful and approved
- Targeting, consent, suppression and privacy requirements are met
- Platform specifications and character limits are current
- Links, tracking parameters and conversion events work
- The output matches the approved brand voice and offer
- A human owner approved customer-facing communication and budget changes
For US campaigns, the FTC maintains current guidance on endorsements, influencers and reviews. Apply the rules of every market where the campaign runs.
Frequently asked questions
What are the best ChatGPT prompts for marketing?
The best prompts use real marketing inputs and produce a specific artifact. Customer-language analysis, campaign briefs, SEO content roadmaps, email sequences, landing-page audits and performance reports are useful because a marketer can review the output against source material and a clear objective.
How do I make ChatGPT marketing output sound less generic?
Provide customer language, approved brand examples, product facts, proof and channel constraints. Ask ChatGPT to preserve factual meaning, mark missing evidence and avoid unsupported claims. Generic output usually means the brief was generic.
Can ChatGPT create a complete marketing strategy?
ChatGPT can organize evidence, compare options and draft a plan. It cannot know your economics, customers, capacity or channel performance unless you provide them. Use it to prepare the decision, then keep accountability with the marketer responsible for results.
Do these marketing prompts work with Claude, Gemini, Grok and Copilot?
Yes. These prompts use model-agnostic instructions and generally work with Claude, Gemini, Grok, Microsoft Copilot and other capable AI assistants. Adjust them for each assistant’s available files, connected tools, web access and context limits.
Can ChatGPT analyze live marketing data?
It can analyze files you provide and, where available and approved, use connected tools or plugins. Give it only the access required for the task, define the source of truth and keep consequential actions behind human approval.
Start with one marketing bottleneck
Pick the prompt closest to the decision blocking your team. Fill every bracket, attach the real source material and review the output against a clear business goal.
For broader operational use cases, see 25 Best ChatGPT Prompts for Business in 2026. Browse the full GPTBot marketing prompt library when you need a more specialized template.
