These 50 ChatGPT prompts for marketers cover four jobs: researching customers and competitors, creating content, planning campaigns, and analyzing results. Use them when you have interview notes to sort, a landing page to draft from an approved brief, or a campaign export that needs to become a decision before the next team meeting.

Choose the template that matches your task, replace the fields in square brackets, and provide the source material you’re authorized to share. Each template specifies what ChatGPT should return. The worked examples show what to check before you use the output in a brief, publish it, or change campaign spend.

ChatGPT Prompts for Marketing Research

Use these templates to turn customer interviews, reviews, sales notes, and competitor pages into evidence you can use in positioning and campaign briefs. Label source material before you paste it so you can trace each finding back to a specific interview, review, or page. Treat gaps in the evidence as questions for your next round of research.

1. Find Themes in Customer Interviews

Analyze [interview notes labeled by participant ID] to answer [research question]. Return a table with theme, short verbatim quote, supporting interview IDs, and number of unique participants who mentioned it. Count each participant once per theme. Include a conflicting example when the notes contain one. Flag missing IDs or unclear quotes and keep those excerpts out of the participant counts.

2. Extract Voice-of-Customer Phrases

Extract exact customer wording from [interview excerpts, reviews, or support notes] for [product and audience]. Group phrases under problems, desired outcomes, and objections. Return a table with the original phrase, source ID, and situation in which the customer used it. Preserve the wording and label unclear context. If a category has no usable quotes, leave it empty and state what evidence is missing.

3. Analyze Customer Reviews by Theme

Analyze [review dataset] for [product or category] during [date range]. Deduplicate by review ID where available. Return each theme, its sentiment, the number of distinct reviews mentioning it, the total number of eligible reviews, and one supporting excerpt. A review can count toward several themes. Flag possible duplicates when IDs are missing and describe frequencies as findings from this sample.

4. Build Evidence-Based Buyer Personas

Use [interview notes, CRM summaries, or survey findings] to draft buyer profiles for [product]. For each profile, list the job the buyer needs done, purchase trigger, decision criteria, objections, and supporting source IDs. Label each point as observed or a hypothesis. Where evidence is too thin to distinguish two profiles, list the questions that would help confirm whether they belong in separate segments.

5. Map Jobs to Be Done

Analyze [customer interview excerpts] for [product category]. For each recurring job, write one statement in the form “When [situation], I want to [make progress], so I can [achieve an outcome].” Add the current workaround and a source ID. Separate the customer’s words from your interpretation. If the situation or outcome is unclear, mark the missing element and suggest one follow-up question.

6. Draft Customer Interview Questions

Write a [interview length]-minute discussion guide for [participant group] to investigate [research objective]. Ask about a recent purchase or attempted purchase, including the trigger, alternatives considered, decision criteria, and what happened afterward. Pair each main question with one neutral follow-up. If the objective includes a hypothesis, convert it into questions about observed behavior and flag any missing recruiting criteria.

7. Audit a Customer Survey for Bias

Review [survey questions and answer choices] against [research objective] for [audience]. Flag leading wording, double-barreled questions, missing options, overlapping ranges, and inconsistent scales. Return the original item, the issue, how it could affect responses, and a revised version. If you need the survey order or screening rules to assess a question, mark that assessment as pending.

8. Extract Objections From Sales Call Notes

Extract buyer objections from [sales call notes with deal IDs and stages]. Group them by sales stage and return the objection, a buyer quote or explicitly labeled paraphrase, supporting deal IDs, and a follow-up question a salesperson could ask. Distinguish objections buyers voiced from concerns the sales team inferred. Put records with missing stages in an “Unknown stage” row.

9. Build a Competitor Comparison Table

Compare [competitors] for [buyer segment] using [source URLs or excerpts] and [comparison criteria]. Return a table with competitor, criterion, finding, supporting URL, source date if available, and verification status. If a page is inaccessible or the source does not establish the finding, write “Not verified.” Record conflicting sources separately. List the claims a marketer should check before publishing the comparison.

Illustrative example. A marketer comparing two approval tools provides public pricing-page excerpts. DraftFlow lists its Team plan at $49 per month, with guided onboarding “available” and no plan named. ProofDesk lists its Team plan at $59 per month and explicitly includes a kickoff call. The comparison row for DraftFlow reads “Team plan onboarding | Not verified | Pricing page mentions onboarding without naming a plan.” The marketer removes the draft claim that only ProofDesk offers onboarding and checks DraftFlow’s plan terms before publishing.

For the full process of collecting sources, choosing comparison criteria, and checking the finished table, see How to Create a Competitor Comparison Table With ChatGPT.

10. Compare Competitor Positioning

Compare how [competitors] position themselves using [homepage or product-page excerpts]. Return a matrix with stated audience, primary promise, supporting proof, and a short quote with its source. Label an audience inferred from the copy as an inference. If the excerpts contain no proof for a promise, write “No proof in supplied material.” Summarize two differences that matter to [buyer segment].

11. Identify Positioning Gaps to Validate

Compare [customer needs and supporting evidence] with [competitor messaging excerpts]. Identify up to five needs that appear underrepresented in the supplied messaging. For each, give the customer evidence, the competitor material reviewed, a positioning hypothesis, and one way to test it. If competitor coverage is incomplete, mark the hypothesis as provisional and name the page or source to check next.

12. Classify Keywords by Search Intent

Group [keyword list] by likely search intent for [market and audience]. Return a table with query cluster, representative keyword, likely intent, funnel stage, recommended page type, and confidence. Use [supplied search result notes] to resolve ambiguous queries. If result notes are missing, label the classification as inferred and list the clusters to validate in search results.

ChatGPT Prompts for Content Marketing

Use these prompts when you have a brief or source material and need a draft for a specific channel. Give ChatGPT the audience, funnel stage, offer, and facts it can use. An article outline should answer the reader’s question, an email should give the reader a reason to act, and a landing page should support its claims with approved evidence.

Give the model a reference for voice and claims. Include one or two approved writing samples and a list of product claims your team has cleared for this audience and channel. Tell ChatGPT which sample to use if they differ in tone.

13. Create an SEO Content Brief

Create a content brief for [primary query], [target reader], and [page goal] using [search result notes and available sources]. Return the likely intent, proposed H1 and H2s, questions each section should answer, original examples to include, and facts that need verification. Define what the reader should be able to do after reading. If search result notes are missing, mark the intent as provisional.

14. Build a Content Calendar by Funnel Stage

Build a [time period] content calendar for [audience and marketing goal] across [channels], using [available assets and team capacity]. Return a table with publication date, funnel stage, topic, format, CTA, and dependencies. Fit the plan to the team’s capacity. If an asset or approval date is unknown, mark it as pending and list the affected slots.

15. Turn an Expert Interview Into an Article Outline

Turn [expert interview transcript] into an article outline for [reader and article goal]. Give each proposed H2 one takeaway, supporting transcript timestamps or excerpt IDs, and an example the reader can apply. Preserve direct quotes verbatim and label paraphrases. Where the transcript offers no support for a proposed section, flag the gap and write one question for the expert.

16. Write Landing Page Copy From a Product Brief

Draft landing page copy for [audience] at [funnel stage], using [product brief], [approved claims], and [primary CTA]. Return a hero headline, supporting line, three benefit blocks with evidence, two objection responses, and CTA copy. Use [brand voice samples] as a style reference. Put any proposed claim that lacks support in a separate “Needs approval” list.

Illustrative example. A brief for a creative approval tool confirms that agency teams can collect comments and approve files in one workspace. The draft hero reads “Keep client feedback and creative approvals in one place,” with the CTA “Book a demo.” ChatGPT also flags “Cut approval time by 50%” as a claim that needs evidence. The marketer keeps the approved workflow claim and removes the percentage from the draft.

17. Draft a Customer Case Study From Approved Notes

Draft a customer case study from [approved interview notes], [implementation timeline], and [verified results]. Use three parts: the customer’s problem, what changed, and the measured outcome. Include the period and baseline for each metric when supplied. Insert “[Confirm with customer]” beside missing details and list quotes, metrics, and names still awaiting approval.

18. Write an Email Nurture Sequence

Write a [number]-email nurture sequence for [audience] after [signup or download trigger]. Use [offer], [approved proof], [brand voice], and [sending cadence]. For each email, provide its objective, subject, body, and one CTA. Give each message a distinct reason to open and act. If timing or eligibility rules are missing, flag those decisions for the campaign owner.

19. Generate Email Subject Line and Preview Text Pairs

Generate [number] subject line and preview text pairs for [email body], [audience], and [campaign objective]. Respect [character limits] and group pairs by the hypothesis they test, such as benefit specificity or question framing. Make the preview text add information. If the email body does not support a proposed claim or deadline, omit it and flag the gap.

20. Adapt One Message for Multiple Social Channels

Adapt [approved core message] for [social channels] using [audience by channel], [format and length limits], and [CTA]. Return a hook, publishable caption draft, and suggested visual for each channel. Preserve the offer terms and product facts. If a required link, visual asset, or channel specification is missing, mark it as a publishing dependency.

21. Repurpose a Webinar Into Content Assets

Use [webinar transcript with timestamps] to produce three short social post drafts, one follow-up email, and three clip ideas for [audience]. For each asset, cite the source timestamp and state its intended CTA. Flag claims that need a source beyond the webinar. If the transcript lacks timestamps, identify clip boundaries by speaker and opening and closing phrases.

22. Write a Short-Form Video Script

Write a [duration]-second video script for [channel, audience, and objective] using [approved message] and [available footage]. Return a shot list with time ranges, visuals, spoken lines, on-screen text, and one CTA. Keep the timing realistic for the voiceover. If a scene requires footage the team lacks, flag the missing shot and suggest an available substitute.

23. Rewrite Copy in Your Brand Voice

Rewrite [draft] for [audience and channel] using [approved brand voice samples]. Preserve product facts, commitments, required wording, and the CTA. Return the revised draft and a short list of edits that might affect meaning. Where the draft and approved samples conflict on a factual claim or term, flag the conflict for review.

24. Refresh an Existing Article for Search Intent

Review [current article] alongside [target queries], [search result notes], and [available performance data]. Return a prioritized edit list showing the section, suggested change, reader need, and evidence. Identify passages worth keeping and facts to recheck. If current result or performance data is missing, flag recommendations that depend on it.

25. Audit Marketing Copy for Unsupported Claims

Audit [copy draft] against [approved claims and supporting documents]. Extract each quantitative, comparative, or outcome claim and return its exact wording, supporting evidence, and status: supported, needs evidence, or conflicts with source. Suggest a revision based on the verified facts. If supporting material is missing entirely, mark the affected claims as pending review.

ChatGPT Prompts for Marketing Campaigns

Start with the campaign goal, audience, offer, budget, dates, and the people who will execute the work. These prompts help you translate those inputs into channel decisions, creative briefs, and launch checks. Give the model your existing constraints, such as a fixed landing page or a single designer, so the plan fits the campaign you can actually run.

26. Turn a Campaign Brief Into a Launch Plan

Turn [campaign brief], [launch date], and [team capacity] into a launch plan. Work backward from the date and return a table with milestone, owner, deadline, dependency, and acceptance criterion. Include approvals for creative, landing pages, and tracking. Mark unassigned owners as “Unassigned” and identify missing decisions that could delay launch.

27. Assign a Role to Each Campaign Channel

Use [campaign goal], [target audiences], [available channels], and [resources] to assign a role to each channel. Return a matrix with channel, audience, funnel role, CTA, and success metric. Identify where one channel hands off to another. If audience reach or tracking capability is unknown, flag the proposed role for validation.

28. Build a Messaging Matrix by Audience Segment

Build a messaging matrix for [audience segments] using [customer research], [offer], and [approved evidence]. Give each segment a buying trigger, primary message, supporting proof, common objection, and CTA. Cite the source for each segment insight. Where two segments share a message or there is insufficient evidence to distinguish them, identify the shared message and the evidence needed to refine it.

29. Develop Paid Ad Angles From Customer Research

Develop [number] paid ad angles for [audience and channel] from [customer interviews or review themes] and [approved offer]. For each angle, provide the source insight, opening hook, proof point, and creative concept. Mark ideas with weak evidence as hypotheses. If an angle depends on an unapproved claim, identify the evidence or approval it needs.

30. Write Search Ad Variants for One Ad Group

Draft [number] search ad variants for [ad group keywords and intent], [landing page copy], and [approved offer]. Use [headline and description character limits] and return a table with headline, description, character counts, and the message each variant emphasizes. Flag any asset that still needs a platform length check. If limits are missing, mark drafts as awaiting length validation.

31. Write Paid Social Ad Copy Variants

Write [number] paid social ad copy variants for [audience, placement, and offer]. Vary [one test variable] and hold [control elements] constant. Return primary text, headline, CTA, and the hypothesis for each version. Use only [approved claims]. If the proposed variable requires changing the offer or audience too, flag the confound and suggest a cleaner test.

32. Create a Creative Brief for Campaign Assets

Write a creative brief from [campaign goal, audience insight, approved message, and placements]. Return deliverables by format, production specs, required elements, available assets, approval owner, and acceptance criteria. Keep the core message consistent across placements. Mark unconfirmed dimensions, usage rights, or production dates as open items.

33. Map Retargeting Messages to Funnel Stages

Map messages to [approved engagement-based segments] for [offer and campaign goal]. For each segment, return the qualifying interaction, likely next question, message, CTA, and exclusion rule. Use only supplied engagement events and permitted audience definitions. If consent, exclusions, or segment eligibility is unclear, mark the segment as pending review before launch.

34. Build a Creative Testing Matrix

Design a creative test for [campaign objective, audience, assets, and budget]. Return a table with hypothesis, control, variant, test variable, primary metric, guardrail, and decision criterion. State which campaign settings need to stay fixed. If baseline performance or expected volume is missing, list what you need to set a credible evaluation threshold.

Illustrative example. A demo campaign tests two opening lines. Version A leads with “Still chasing approvals by email?” Version B leads with “Can your team see what’s ready to publish?” Both use the same visual, audience, offer, and landing page. The test matrix lists opening message as the variable and cost per qualified demo request as the primary metric. The marketer waits until enough requests have been reviewed under the same qualification rules before comparing the variants.

To connect the campaign brief, creative variants, and performance review into a repeatable process, see ChatGPT for Marketing: 15 Workflows With Prompts.

35. Compare Campaign Budget Scenarios

Compare [proposed budget allocations] using [historical costs and conversion data], [budget cap], and [campaign goal]. Return a scenario table with spend by channel, projected results, formulas, and key assumptions. Show how results change when [cost or conversion assumption] varies. Flag assumptions about performance at higher spend and leave projections blank where required inputs are missing.

36. Define Campaign UTMs and Tracking Checks

Generate a tracking plan for [destination URLs, channels, and creative variants] using [UTM naming conventions]. Return a table with destination, UTM parameters, final tagged URL, expected analytics values, and a prelaunch check. Preserve existing URL parameters and flag conflicts. If a destination or naming rule is missing, mark the link as pending.

37. Run a Pre-Launch Campaign Review

Review [launch plan], [creative], [landing page details], [tracking plan], and [approval status] against [launch criteria]. Return issues grouped by launch blocker, fix before launch, and monitor after launch. For each issue, specify the affected asset, owner, and verification step. Mark any item with missing evidence as “Unchecked.”

ChatGPT Prompts for Marketing Analytics

Use these templates with an approved export or a table you can reconcile with your reporting system. Provide the date range, currency, metric definitions, and attribution window. Ask for the inputs and formula behind each calculation. For interpretation, require the output to distinguish what the data shows from a possible explanation that still needs testing.

Check comparability first. Confirm that the rows use the same date boundaries, currency, conversion definitions, and attribution rules. Define how to handle missing values and zero denominators. Comparing a seven-day click-through report with a report that includes view-through conversions can change the apparent ranking of your channels.

38. Define KPIs for a Campaign Goal

Recommend one primary KPI and up to three supporting metrics for [campaign goal, funnel stage, and decision]. Use [available data sources and tracking events]. For each metric, specify its definition, formula, data source, reporting cadence, and owner. If the necessary event is unavailable, state the tracking gap and label any proposed proxy.

39. Check a Campaign Export for Data Quality Issues

Audit [campaign export] using [data dictionary, expected date range, and currency]. Check for duplicate keys, missing values, date gaps, inconsistent currency, unexpected totals, and incompatible metric definitions. Return the affected rows or fields, the likely impact, and the check needed to resolve each issue. Flag tests that require a field or definition missing from the export.

40. Calculate CTR, CPC, CPA, and ROAS

Calculate CTR as clicks divided by impressions, CPC as spend divided by clicks, CPA as spend divided by [defined actions], and ROAS as attributed revenue divided by spend for [campaign table]. Report CTR as a percentage and ROAS as a multiple. Show raw inputs and formulas for each row, plus totals calculated from summed inputs. Mark missing inputs or zero denominators as “N/A” and explain why.

41. Compare Campaign Performance Across Channels

Compare [channel reports] for [date range and objective] using [shared metric definitions and attribution rules]. Return each channel’s spend, eligible conversion count, and comparable cost metric. Put platform-reported metrics with incompatible attribution in a separate column and explain the limitation. If common conversion data is missing, state which rankings the available data can support.

42. Compare Marketing Performance Period Over Period

Compare [current period data] with [prior period data] for [metrics]. Show both date ranges, raw values, absolute changes, and percentage changes where the prior value is nonzero. For comparable volume metrics, include per-day figures if the periods differ in length. Flag partial periods, tracking changes, and any metric that cannot be compared on the supplied definitions.

43. Find the Biggest Drop-Off in a Marketing Funnel

Analyze [ordered funnel steps and counts] for [cohort and date range]. Calculate step-to-step conversion and absolute losses using the previous step as the denominator. Identify the largest percentage drop and largest numerical loss. List two hypotheses and a check for each. If counts mix cohorts or event definitions, flag the affected steps before interpreting them.

44. Diagnose a Sudden Increase in CPA

Investigate the change in CPA between [baseline period] and [comparison period] using [spend, clicks, conversions, attribution settings, and change log]. Show how spend, CPC, and click-to-conversion rate moved. Return three columns: observed change, possible explanation, and verification step. If recent conversions may still arrive or the action definition changed, state which conclusions are premature.

45. Segment Performance by Device or Geography

Break down [campaign data] by [device or geography] for [date range]. Show spend, traffic, conversion counts, and conversion rates using [conversion denominator]. Flag groups below [minimum sample threshold] and show the counts behind every rate. If no threshold or denominator was supplied, flag the missing definition before drawing conclusions about segment performance.

46. Review an A/B Test Before Calling a Winner

Review [experiment design, group sizes, conversions, dates, and decision rule]. Return the observed effect, sample sizes, and any issues with assignment, exposure, tracking, or early stopping. Use [preselected analysis method] if the data supports it. If the planned sample or analysis requirements are missing, report the descriptive result and identify what is needed to make a decision.

47. Audit an Attribution Report for Misleading Comparisons

Audit [platform reports and first-party report] using [attribution windows, models, and conversion definitions]. Identify differences in credit rules, click-through and view-through inclusion, duplicated conversion credit, and reporting dates. Return each affected comparison, the reason it may mislead, and a way to reconcile it. If report settings are unavailable, mark the comparison as unverified.

48. Analyze Lead Quality by Source

Compare lead sources using [source-level spend, lead count, qualified lead count, date range, and qualification rules]. Return cost per lead, qualification rate, and cost per qualified lead with formulas and inputs. Include pending lead reviews when available. Use “N/A” for zero denominators and flag sources with inconsistent qualification rules or different amounts of time for leads to mature.

Illustrative example. Channel A spent $1,000 and generated 100 leads, including 10 qualified leads. Channel B spent $900 and generated 45 leads, including 15 qualified leads. The output shows a cost per lead of $10 for A ($1,000 ÷ 100) and $20 for B ($900 ÷ 45). Cost per qualified lead is $100 for A ($1,000 ÷ 10) and $60 for B ($900 ÷ 15). Before moving budget, the marketer confirms that both sources use the same qualification rules and checks how many leads are still awaiting review.

49. Write a Weekly Marketing Performance Summary

Write a weekly update for [stakeholders] using [this week’s metrics, prior-week metrics, targets, and change log]. Lead with the decision that needs attention. Then list three material changes with figures, possible explanations, and next checks with owners. Distinguish measured changes from hypotheses. If a metric is missing or the period is incomplete, flag it beside the affected conclusion.

50. Turn a Campaign Postmortem Into Next Actions

Turn [campaign postmortem], [planned versus actual results], and [team constraints] into a prioritized action list. For each action, include the supporting finding, proposed owner, deadline, and success check. Separate changes the evidence supports from ideas that need another test. If the postmortem leaves a major outcome unexplained, add a verification task before proposing a corrective action.

Limits and Risks of Using ChatGPT for Marketing

  • Invented facts and competitor claims. Open the cited source and check that it supports the exact claim. A competitor can list a feature on its product page without offering it on the plan you’re comparing. Verify prices, plan terms, quotes, and case study figures in the original material.
  • Confidential data exposure. Use a workspace your organization has approved for the material. Remove names, email addresses, account identifiers, and unnecessary free-text fields from exports. For channel analysis, aggregated counts often provide enough detail.
  • Incorrect calculations and misleading comparisons. Recalculate decision-critical figures in your spreadsheet or reporting system. Check denominators, date ranges, and attribution windows. For example, calculate total CPA as total spend divided by total eligible actions. An average of campaign CPAs gives small campaigns the same weight as large ones.
  • Confident explanations with weak evidence. Suppose CPA rose from $50 to $65 and the reported click-to-conversion rate fell. Those are observations. A broken form is one possible explanation. Check the form on affected devices, review tracking changes, and allow for conversion reporting delays before naming a cause.

What Marketers Must Review Before Acting

The editor or product owner confirms that claims, quotes, and offer terms are accurate before publication. The analyst checks the calculations and decides whether the underlying reports are comparable before presenting a finding. The campaign owner approves budget changes, launch decisions, and calls to end a test.

Pick one task you already have on your desk. Run the matching template with source material you can verify, then check the part of the answer that will drive your next decision. For landing page copy, that might be one product claim. For a budget review, it might be the definition and count of qualified leads.

FAQ

What are the best ChatGPT prompts for marketers?

Start with prompts tied to a specific decision or deliverable, such as grouping customer interview themes, drafting a landing page from an approved brief, designing an ad test, or comparing qualified lead costs. Include your source material and specify the output you need.

How do I write a good ChatGPT prompt for marketing?

State the task, audience, goal, source material, constraints, and output format. For example, ask for three landing page headlines for a defined segment, based on approved product claims, with one distinct value proposition per headline.

Can ChatGPT help with marketing research?

Yes. It can organize interview notes, group customer reviews, and compare supplied competitor pages. Check quotes and claims against the original sources, and use additional research to validate patterns drawn from small samples.

Can ChatGPT analyze marketing campaign data?

Yes, if you provide an approved export or paste a table with clear column definitions and date ranges. Ask it to show its formulas and inputs, then verify the calculations before sharing results or changing spend.

Can I use ChatGPT to write marketing emails and ads?

Yes. Give it the audience, offer, channel, tone, and approved claims. Review each draft for accuracy, brand fit, and applicable advertising or platform requirements before it goes live.

What marketing data should I avoid uploading to ChatGPT?

Keep customer identifiers, raw CRM records, confidential contracts, and unreleased campaign information out of unapproved tools. Use an approved workspace and share only the fields needed for the task.