30 Claude Prompts for Marketing That Drive Results (2026)
Most marketers who try Claude type a one-line prompt, get a generic answer, and blame the model. The model is fine. The prompts are wrong. A good Claude prompt for marketing is specific, scoped, and tells Claude exactly what you'd tell a capable new hire on day one.
This post gives you 30 of those. Organized by the work you actually do — SEO, ads, email, social, analytics, strategy — each one copy-pasteable, each one tested. If you want to skip ahead and build your own, use the free Claude prompt generator and drop in your specifics.
TL;DR
- 30 Claude marketing prompts organized by function: SEO, ads, email, social, analytics, strategy.
- Each prompt is free, ungated, and copy-pasteable. Swap
[bracketed placeholders]with your own values. - Use Claude Opus 4.7 or Sonnet 4.6 for best results on long-context marketing work.
- To customize any prompt for your brand, paste it into the Claude prompt generator and add your context.
Table of Contents
- How to use these Claude marketing prompts
- SEO & content prompts (5)
- Ads & paid media prompts (5)
- Email & lifecycle prompts (5)
- Social & community prompts (5)
- Analytics & reporting prompts (5)
- Strategy & positioning prompts (5)
- Pro tips for marketing with Claude
- FAQ
How to use these Claude marketing prompts
Every prompt below follows the same structure: a role, the task, context, constraints, and the output shape. This mirrors what Anthropic recommends in their prompt engineering guidance — be explicit, specific, and give Claude what it needs to get it right the first time.
Three ways to use each prompt:
- Copy and paste. Swap the
[brackets]with your values and send it to Claude directly. - Customize in the generator. Paste the prompt into the Claude prompt generator, add your brand voice and constraints, and it outputs a production-ready version.
- Save as a Claude Project. For prompts you'll reuse weekly (reporting, content briefs), drop them in a Claude Project with your brand context in the project instructions.
For the full categorized hub, see /prompts/marketing/claude. For SEO-specific Claude workflows, see the Claude SEO prompts hub.
SEO & content prompts
SEO and content is where Claude shines — long-context reasoning, tone control, and structured outputs matter here more than anywhere else in marketing.
1. Keyword cluster builder
What it does: Groups a raw keyword list into topic clusters mapped to content pages.
Prompt:
You are an SEO strategist. I will paste a list of keywords from a keyword research tool.
TASK:
Group them into topical clusters. For each cluster, suggest one pillar page and three supporting posts.
CONTEXT:
- Niche: [YOUR NICHE, e.g. "AI prompt engineering tools"]
- Primary audience: [e.g. "marketers and solo SaaS founders"]
- Existing pages we already have: [LIST URLs or topics]
CONSTRAINTS:
- Cluster by search intent, not just word overlap.
- Ignore branded competitor keywords.
- Flag any keyword you consider low commercial value.
OUTPUT FORMAT:
| Cluster name | Pillar page topic | Supporting posts (3) | Primary intent | Notes |
KEYWORDS:
[PASTE KEYWORDS]
Swap: [YOUR NICHE], [PRIMARY AUDIENCE], [EXISTING PAGES], [PASTE KEYWORDS].
Tip: Feed this into Claude after pulling 300–500 keywords from Ahrefs or a GSC export. Claude Sonnet 4.6 handles this cleanly up to ~1,500 keywords per pass.
2. Content brief generator
What it does: Turns a target keyword into a publish-ready content brief.
Prompt:
You are a senior content strategist.
TASK:
Write a complete content brief for a blog post targeting the keyword: "[TARGET KEYWORD]"
CONTEXT:
- Our site: [YOUR DOMAIN]
- Existing authority topics: [2-3 TOPICS]
- Reader persona: [ONE SENTENCE ON WHO READS THIS]
CONSTRAINTS:
- Length recommendation must match the SERP for this keyword.
- Use search intent (informational / commercial / transactional) to shape the angle.
- Flag 3 unique angles competitors are missing.
OUTPUT FORMAT:
- Suggested H1 (under 60 chars)
- Meta description (under 155 chars)
- Reader problem in one sentence
- Full H2 outline (6-10 sections with notes)
- 5 FAQ questions worth answering
- 3 internal link opportunities (generic descriptions)
- 3 unique angles vs. top-ranking competitors
Swap: [TARGET KEYWORD], [YOUR DOMAIN], [EXISTING AUTHORITY TOPICS], [READER PERSONA].
Tip: Paste the top 3 SERP URLs into the same prompt for sharper competitor analysis.
3. Meta description rewriter (CTR fix)
What it does: Rewrites a page's title + meta to improve click-through rate.
Prompt:
You are a CRO-minded SEO writer.
TASK:
Rewrite the title tag and meta description for a page that ranks but doesn't get clicked.
CURRENT STATE:
- Page URL: [URL]
- Current title: [CURRENT TITLE]
- Current meta: [CURRENT META]
- Target keyword: [KEYWORD]
- Current CTR: [CTR]% at position [POSITION]
- Audience: [WHO]
CONSTRAINTS:
- Title ≤ 60 characters, keyword near the front.
- Meta ≤ 155 characters, ends with a soft CTA.
- Do not lie or imply content the page doesn't deliver.
- Match commercial or informational intent based on the keyword.
OUTPUT:
- 5 title options (ranked by predicted CTR)
- 3 meta descriptions (ranked by predicted CTR)
- One-line reasoning for the top pick
Swap: [URL], [CURRENT TITLE], [CURRENT META], [KEYWORD], [CTR], [POSITION], [WHO].
Tip: Run this on any page ranking in positions 4–10 with CTR under 3%. These are free-click opportunities.
4. Internal link audit
What it does: Recommends internal links to pass equity to underperforming pages.
Prompt:
You are a technical SEO doing an internal linking audit.
TASK:
Suggest internal link opportunities to strengthen this target page.
TARGET PAGE:
- URL: [URL]
- Topic: [TOPIC]
- Why it matters: [WHY — e.g. "ranks at pos 12 for a high-value keyword"]
DONOR PAGES (pages already ranking well on related topics):
[LIST URLS + TOPIC FOR EACH]
CONSTRAINTS:
- Suggest anchor text that uses the target keyword or a close variant.
- Anchors should read naturally, not spammy.
- Flag any donor page where adding the link would disrupt reading flow.
OUTPUT:
| Donor URL | Exact sentence to insert/modify | Suggested anchor text |
Swap: [URL], [TOPIC], [WHY], [DONOR PAGES LIST].
Tip: For best results, paste the full text of each donor page into the context block. Claude's long context handles this easily.
5. Content refresh plan
What it does: Audits an older blog post and returns a prioritized refresh plan.
Prompt:
You are an SEO editor.
TASK:
Audit the pasted blog post and output a refresh plan to improve rankings.
CONTEXT:
- Target keyword: [KEYWORD]
- Current position: [POSITION]
- Published: [DATE]
- Competing top-3 URLs: [URLS]
CONSTRAINTS:
- Keep what's already working.
- Flag stale facts, dates, stats, product references, or outdated model names.
- Suggest new H2s if competitors cover topics we miss.
OUTPUT:
- Verdict: keep / refresh / rewrite
- 5 top-priority changes (ranked by impact)
- 3 new H2s to add
- Stats/dates that need updating (line-referenced)
- Suggested new title + meta if current ones underperform
POST:
[PASTE FULL BLOG POST]
Swap: [KEYWORD], [POSITION], [DATE], [URLS], [PASTE FULL BLOG POST].
Tip: Run this quarterly against any post that's dropped more than 5 positions in 90 days.
Faster version: Generate a custom SEO prompt in 10 seconds →
Ads & paid media prompts
Claude is unusually strong at ad copy because it resists the generic AI tone that tanks CTR. The key is giving it constraints that match your audience.
6. Ad copy variant generator
What it does: Produces 10 on-brand ad copy variants for A/B testing.
Prompt:
You are a performance marketing copywriter.
TASK:
Write 10 ad copy variants for the product below, ready to test.
PRODUCT:
- Name: [PRODUCT]
- Offer: [OFFER]
- Audience: [AUDIENCE]
- Main pain we remove: [PAIN]
- Main gain we deliver: [GAIN]
- Proof point: [STAT, TESTIMONIAL, OR CREDENTIAL]
CHANNEL:
[Google Search / Meta / LinkedIn / X]
CONSTRAINTS:
- Match the character limits of the channel above.
- No emoji unless the brand voice uses them.
- Avoid vague claims ("next-gen", "cutting-edge").
- Each variant should test a different angle (pain, status, time, money, risk-reversal, social proof, curiosity, authority, contrast, outcome).
OUTPUT:
| # | Angle | Headline | Body | CTA |
Swap: [PRODUCT], [OFFER], [AUDIENCE], [PAIN], [GAIN], [PROOF POINT], [CHANNEL].
Tip: Give Claude three examples of your best-performing past ads in the prompt — it'll match voice far more accurately.
7. Landing page angle generator
What it does: Produces 5 distinct landing page angles for the same offer.
Prompt:
You are a direct-response strategist.
TASK:
Propose 5 distinct landing page angles for the same offer. Each angle = a different emotional entry point.
OFFER:
[ONE PARAGRAPH DESCRIBING THE OFFER]
AUDIENCE:
[WHO IS BUYING]
CONSTRAINTS:
- Each angle must have a different hook, not just different words.
- Angles should span: status, outcome, fear-of-loss, curiosity, contrast.
- Keep every angle truthful — no fabricated claims.
OUTPUT PER ANGLE:
- Angle name
- Above-the-fold H1
- Subhead
- 3 bullet pains it addresses
- One-line social proof format
- Primary CTA
Swap: [OFFER], [AUDIENCE].
Tip: Use this before design. Pick the two strongest angles and A/B test them as separate pages.
8. Audience brief for paid media
What it does: Generates a deep audience brief that media buyers can act on.
Prompt:
You are a paid media strategist.
TASK:
Write an audience brief for targeting the persona below on paid channels.
PERSONA:
- Role: [ROLE]
- Company size: [SIZE]
- Budget authority: [YES/NO/INFLUENCER]
- Top 3 daily pains: [LIST]
OUR PRODUCT:
[ONE PARAGRAPH]
CONSTRAINTS:
- Infer likely content consumption (YouTube channels, podcasts, newsletters, subreddits).
- Propose 5 targeting angles per platform (Meta, LinkedIn, Google).
- Flag any segment that's likely to be unprofitable and why.
OUTPUT:
- Persona summary (3 sentences)
- Platform-by-platform targeting angles
- Objections to address in creative
- 3 messaging tests worth running
Swap: [ROLE], [SIZE], [BUDGET AUTHORITY], [DAILY PAINS], [PRODUCT DESCRIPTION].
Tip: Feed past customer interview transcripts in as context — you'll get a brief grounded in real voice, not guesses.
9. Creative brief for a campaign
What it does: Produces a one-page creative brief designers can execute against.
Prompt:
You are a creative director.
TASK:
Write a one-page creative brief for the campaign below.
CAMPAIGN:
- Goal: [MEASURABLE GOAL]
- Audience: [AUDIENCE]
- Product: [PRODUCT]
- Timeline: [WEEKS]
- Budget: [BUDGET]
CONSTRAINTS:
- Brief must be readable in under 3 minutes.
- Avoid marketing jargon.
- Every section must answer a specific question a designer or copywriter would ask.
OUTPUT SECTIONS:
1. One-sentence campaign summary
2. The single insight this campaign is built on
3. Audience + what they believe today vs. what we want them to believe
4. Mandatory proof points (stats, testimonials, logos)
5. Tone: 3 words that describe voice, 3 words that describe what to avoid
6. Deliverables checklist
7. Success metric + threshold
Swap: [MEASURABLE GOAL], [AUDIENCE], [PRODUCT], [WEEKS], [BUDGET].
Tip: Save the output as your campaign kickoff doc. Drop it in a Claude Project so the whole team can iterate against the same brief.
10. A/B test hypothesis generator
What it does: Generates 10 testable hypotheses for a landing page or ad set.
Prompt:
You are a CRO analyst.
TASK:
Generate 10 A/B test hypotheses for the asset below, ranked by expected impact.
ASSET:
- Type: [landing page / ad / email]
- Current state: [PASTE COPY OR URL]
- Current conversion rate: [RATE]
- Bottleneck: [WHERE USERS DROP, IF KNOWN]
CONSTRAINTS:
- Each hypothesis must be one sentence in "if we change X, then Y will happen because Z" form.
- No hypothesis that requires rebuilding the whole asset.
- Flag hypotheses that require dev work vs. copy-only changes.
OUTPUT:
| # | Hypothesis | Effort (copy-only / design / dev) | Expected lift | Why |
Swap: [TYPE], [ASSET COPY OR URL], [CURRENT CONVERSION RATE], [BOTTLENECK].
Tip: Pair this with a heatmap export or session replay notes for far sharper hypotheses.
Email & lifecycle prompts
Email is the most leveraged marketing channel and the one most hurt by generic AI copy. Specificity is everything.
11. Welcome sequence (5 emails)
What it does: Writes a complete 5-email welcome sequence from scratch.
Prompt:
You are an email marketer specialized in lifecycle campaigns.
TASK:
Write a 5-email welcome sequence for new signups.
PRODUCT:
[ONE PARAGRAPH]
AUDIENCE:
[WHO SIGNED UP]
ACTIVATION GOAL:
[THE KEY ACTION WE WANT THEM TO TAKE]
CONSTRAINTS:
- Each email must drive toward activation, not just introduce the brand.
- Subject lines under 50 characters.
- Each email under 150 words of body copy.
- No "hope you're well" openers.
OUTPUT PER EMAIL:
- Send delay (Day 0, Day 1, Day 3, Day 5, Day 7)
- Purpose in one sentence
- Subject line (plus 2 variants)
- Preview text
- Body copy
- Single CTA
Swap: [PRODUCT], [AUDIENCE], [ACTIVATION GOAL].
Tip: Paste your best-performing past welcome email as a tone reference — Claude will match it.
12. Re-engagement email
What it does: Writes a win-back email for users who went dark.
Prompt:
You are an email marketer.
TASK:
Write a re-engagement email to subscribers who haven't opened in 60+ days.
CONTEXT:
- Product: [PRODUCT]
- What's new since they last engaged: [1-3 ITEMS]
- Goal: get them to click once, not to sell
CONSTRAINTS:
- Subject line must be pattern-interrupting but honest.
- Body under 100 words.
- Single CTA.
- Include a clear "if you don't click, we'll unsubscribe you" note at the bottom (compliance-safe).
OUTPUT:
- 5 subject line variants
- Preview text
- Body copy
- CTA button text
Swap: [PRODUCT], [WHAT'S NEW].
Tip: Send this to dormant segments before a major launch — cleans the list and recovers a surprising percentage.
13. Drip logic planner
What it does: Maps out a behavior-triggered drip campaign.
Prompt:
You are a lifecycle marketing strategist.
TASK:
Design a behavior-triggered drip campaign for the scenario below.
SCENARIO:
- Trigger event: [EVENT, e.g. "user viewed pricing but didn't sign up"]
- Product: [PRODUCT]
- Desired outcome: [CONVERSION GOAL]
- Audience segment: [SEGMENT]
CONSTRAINTS:
- Map out up to 7 touchpoints.
- Include branching logic (if they click X, send Y; if they ignore, send Z).
- Recommend channel per touchpoint (email / SMS / in-app / retargeting).
OUTPUT:
- Campaign flow diagram (text-based tree)
- Per touchpoint: delay, channel, purpose, CTA
- Exit conditions
- Success metrics per stage
Swap: [TRIGGER EVENT], [PRODUCT], [CONVERSION GOAL], [SEGMENT].
Tip: Export the output as a flowchart with Mermaid — ask Claude to add the Mermaid syntax and paste into Notion.
14. Subject line variant generator
What it does: Produces 20 subject line variants across 4 angle types.
Prompt:
You are an email copywriter optimizing for open rate.
TASK:
Write 20 subject line variants for the email below.
EMAIL PURPOSE:
[ONE SENTENCE]
EMAIL CONTENT:
[PASTE OR SUMMARIZE]
AUDIENCE:
[WHO RECEIVES THIS]
CONSTRAINTS:
- 5 curiosity-driven
- 5 value-driven (specific benefit)
- 5 urgency-driven (honest, not fake)
- 5 personalized-feeling (uses a personal angle without actual merge fields)
- All under 50 characters.
- No clickbait that overstates the content.
OUTPUT:
| Angle | Subject Line | Character Count | Predicted Open Rate Lift |
Swap: [EMAIL PURPOSE], [EMAIL CONTENT], [AUDIENCE].
Tip: Send the top-3 by predicted lift to a subject line test. Claude tends to rate conservatively, so even its top picks are worth testing.
15. Product launch announcement email
What it does: Writes a launch email that drives clicks, not just awareness.
Prompt:
You are an email copywriter writing a product launch announcement.
TASK:
Write the launch email.
WHAT'S LAUNCHING:
[PRODUCT/FEATURE + ONE-LINE DESCRIPTION]
WHY IT MATTERS:
[SPECIFIC PROBLEM IT SOLVES]
AUDIENCE:
[CURRENT USERS / WAITLIST / COLD LIST]
CONSTRAINTS:
- Lead with the problem, not the product.
- One CTA.
- Under 200 words.
- No "we're excited to announce" opener.
- Include one specific, falsifiable claim (stat, before/after, etc.).
OUTPUT:
- Subject line (plus 3 variants)
- Preview text
- Body copy
- CTA button text
- P.S. line (optional, use if it sharpens urgency)
Swap: [PRODUCT/FEATURE], [SPECIFIC PROBLEM], [AUDIENCE].
Tip: Ban the phrase "we're excited to announce" — it tanks open-to-click rate. Claude will write around it if you say so explicitly.
Social & community prompts
Social copy is where generic AI gets caught immediately. These prompts force Claude to match your voice rather than default to LinkedIn-template-ese.
16. LinkedIn hook idea generator
What it does: Produces 10 LinkedIn hooks matched to your audience.
Prompt:
You are a B2B content strategist writing for LinkedIn.
TASK:
Generate 10 LinkedIn hooks for the topic below. Each hook is the first 2 lines of a post.
TOPIC:
[TOPIC]
AUDIENCE:
[WHO YOU'RE WRITING TO]
CONSTRAINTS:
- No "I got fired/rejected/laid off" bait.
- Each hook must promise something specific.
- 5 should use a contrarian angle, 5 should use a specific number or stat.
- Target 8th-grade reading level.
OUTPUT:
| # | Hook (2 lines) | Angle | Why it works |
Swap: [TOPIC], [AUDIENCE].
Tip: LinkedIn rewards specificity. Claude writes tighter hooks when you give it one of your best past posts as a reference.
17. LinkedIn thread draft
What it does: Expands a hook into a full LinkedIn post.
Prompt:
You are a LinkedIn ghostwriter for [ROLE, e.g. "a B2B SaaS founder"].
TASK:
Write a complete LinkedIn post from the hook below.
HOOK:
[PASTE HOOK]
CONTEXT:
- Author's expertise: [1 SENTENCE]
- Audience: [WHO]
- Recent post performance insight: [e.g. "short posts do better than long"]
CONSTRAINTS:
- Under 1,300 characters (LinkedIn's "see more" cutoff rewards shorter posts in 2026).
- Line breaks every 1-2 sentences.
- One clear takeaway.
- CTA that invites comments, not clicks.
- No emojis unless brand voice uses them.
OUTPUT:
- Finished post
- 3 alternate closers
- 2 engagement-question variants for the CTA
Swap: [ROLE], [HOOK], [AUTHOR'S EXPERTISE], [AUDIENCE], [PERFORMANCE INSIGHT].
Tip: For personal-brand content, paste 2–3 past posts in the context window. Claude matches the rhythm better than any style-tag approach.
18. Caption variants for Instagram/TikTok
What it does: Produces platform-native captions for a single piece of content.
Prompt:
You are a short-form social writer.
TASK:
Write captions for the post below, optimized separately for Instagram and TikTok.
POST CONTENT:
[DESCRIBE THE VIDEO/IMAGE AND ITS KEY MOMENT]
BRAND VOICE:
[3 WORDS]
AUDIENCE:
[WHO]
CONSTRAINTS:
- Instagram: up to 150 characters, hook in first 8 words.
- TikTok: up to 100 characters, front-load the payoff.
- No hashtag stuffing.
- Suggest 3-5 hashtags per platform.
OUTPUT:
- 3 Instagram caption variants + hashtags
- 3 TikTok caption variants + hashtags
- One-line note on which style to test first
Swap: [POST CONTENT], [BRAND VOICE], [AUDIENCE].
Tip: If you have transcripts of your 10 best-performing videos, include them. Claude catches tonal patterns invisible to templates.
19. Comment reply generator
What it does: Generates on-brand replies to common comment types.
Prompt:
You are a community manager.
TASK:
Write replies to the comment types below. Match our brand voice.
BRAND VOICE:
[3 ADJECTIVES + 1 EXAMPLE LINE]
COMMENT TYPES TO HANDLE:
1. Question about pricing
2. Critical / negative
3. Genuine compliment
4. Competitor comparison
5. Off-topic / spam-adjacent
CONSTRAINTS:
- Each reply under 40 words.
- Never defensive on criticism — ask, don't argue.
- Never link out in the first reply; invite to DM if needed.
- Never say "great question!"
OUTPUT:
| Comment type | Reply | Follow-up if they respond |
Swap: [BRAND VOICE], [EXAMPLE LINE].
Tip: Save these as saved replies in your social tool. Update quarterly so voice stays current.
20. UGC request prompt
What it does: Writes outreach DMs/emails to request user-generated content.
Prompt:
You are a partnerships manager.
TASK:
Write a UGC request message to send to power users.
CONTEXT:
- Product: [PRODUCT]
- User's history with us: [e.g. "6 months, uses daily, mentioned us once on X"]
- What we want them to create: [e.g. "30-sec video showing how they use feature X"]
- What we'll offer in return: [e.g. "$100 gift card + feature on our channel"]
CONSTRAINTS:
- Under 120 words.
- No fake personalization.
- Lead with why we asked *them* specifically.
- Make the ask easy (provide examples, offer to draft a script).
OUTPUT:
- DM version (warmer, shorter)
- Email version (slightly longer, more detail)
- 3 subject lines for the email
Swap: [PRODUCT], [USER'S HISTORY], [WHAT TO CREATE], [OFFER].
Tip: Reference the specific thing this user did (post, comment, feature use). Generic outreach has a <2% response rate; specific outreach hits 15%+.
Analytics & reporting prompts
Claude excels at turning raw data into readable narratives — but only if you feed it clean context and strict format requirements.
21. GA4 insight summary
What it does: Converts a GA4 data export into a 5-bullet executive summary.
Prompt:
You are a senior analyst writing for a non-technical CEO.
TASK:
Summarize the GA4 data below in 5 bullets the CEO can skim in 30 seconds.
DATA:
[PASTE GA4 CSV OR TABLE]
TIMEFRAME:
[e.g. "March 2026 vs. February 2026"]
CONTEXT:
- Business goal: [PRIMARY KPI]
- Recent marketing changes: [LIST]
CONSTRAINTS:
- Every bullet must include a number and its direction (up/down) vs. the previous period.
- No "things are trending well" without a number.
- Flag one thing that's bad news, clearly.
- End with the single recommended action.
OUTPUT:
- 5 bullets (headline + one supporting number each)
- 1 recommended action
- 1 thing to investigate next
Swap: [GA4 DATA], [TIMEFRAME], [PRIMARY KPI], [RECENT MARKETING CHANGES].
Tip: Export GA4 data as CSV, paste directly. Claude handles tables natively — no conversion needed.
22. Executive dashboard narrative
What it does: Writes an executive-ready narrative from a dashboard screenshot or data dump.
Prompt:
You are a business analyst writing the narrative section of a weekly exec dashboard.
TASK:
Turn the data below into a one-page narrative for the exec team.
DATA:
[PASTE ALL KPI NUMBERS — current week, previous week, target]
CONTEXT:
- Our quarter goal: [GOAL]
- Known events affecting the week: [LIST, e.g. "launched new pricing page"]
CONSTRAINTS:
- Open with the single most important finding.
- Group findings by: wins / misses / questions.
- Every claim needs a number.
- End with 3 asks the exec team needs to decide on.
OUTPUT:
- TL;DR (3 lines)
- Wins (3)
- Misses (3)
- Questions to investigate (3)
- Asks for the exec team (3)
Swap: [KPI DATA], [GOAL], [KNOWN EVENTS].
Tip: Run this weekly. Claude Projects with a saved template cuts the time to ~5 minutes per report.
23. SQL-to-insight translator
What it does: Converts SQL query results into plain-English marketing insights.
Prompt:
You are a marketing analyst translating data for non-technical stakeholders.
TASK:
Explain what the SQL result below means for our marketing team.
QUERY PURPOSE:
[ONE SENTENCE]
SQL USED:
[PASTE SQL]
RESULT:
[PASTE RESULT TABLE]
CONSTRAINTS:
- Plain English only — no "cohort", "attribution", or "funnel" without defining it.
- Every claim must cite a specific row/value from the result.
- End with 2 questions this data raises.
OUTPUT:
- 1-sentence TL;DR
- 3 key insights
- 2 follow-up questions
- 1 recommended action
Swap: [QUERY PURPOSE], [SQL QUERY], [RESULT].
Tip: If the SQL result is long, paste just the first 50 rows + the aggregate row.
24. Cohort story writer
What it does: Writes the "story" behind a cohort retention chart.
Prompt:
You are a growth analyst.
TASK:
Describe the cohort retention data below as a story — what likely happened and why.
COHORT DATA:
[PASTE COHORT TABLE — months or weeks]
CONTEXT:
- Product: [PRODUCT]
- Known shipped changes during the period: [LIST WITH DATES]
- Key lifecycle events: [e.g. "pricing change on Feb 1"]
CONSTRAINTS:
- Do not over-attribute — say "likely" and "possibly" when causation isn't provable.
- Call out any cohort that's clearly better or worse and connect it to a known change.
- Suggest the single most leveraged next experiment.
OUTPUT:
- 3-paragraph narrative
- Top-performing cohort + hypothesis
- Worst-performing cohort + hypothesis
- Recommended next experiment
Swap: [COHORT DATA], [PRODUCT], [SHIPPED CHANGES], [LIFECYCLE EVENTS].
Tip: Pair this with the GA4 insight prompt (#21) for a full-picture weekly report.
25. Weekly marketing report template
What it does: Generates a reusable weekly marketing report template you can fill in with data.
Prompt:
You are a marketing ops lead.
TASK:
Design a weekly marketing report template our team can fill in every Friday.
CONTEXT:
- Team size: [SIZE]
- Channels we run: [LIST]
- Primary KPI: [KPI]
- Secondary KPIs: [LIST]
CONSTRAINTS:
- Report must be readable in under 5 minutes.
- Structured so execs can skim and ICs can drill in.
- Include a section for qualitative learnings, not just numbers.
OUTPUT:
- Full template with section headers, placeholders, and short example sentences
- 1-sentence description of why each section exists
- Suggested cadence adjustments (e.g. what to report monthly vs. weekly)
Swap: [TEAM SIZE], [CHANNELS], [PRIMARY KPI], [SECONDARY KPIS].
Tip: Paste it into Notion as a template. Your team fills a new page each week — and Claude can draft the narrative section from the raw numbers.
Save time on analytics prompts: customize any of these in the Claude prompt generator →
Strategy & positioning prompts
These prompts do what juniors often can't: structured thinking under pressure. Use them as a starting draft, not a final deliverable.
26. ICP refinement
What it does: Sharpens a fuzzy ICP into one that sales and marketing can actually use.
Prompt:
You are a B2B positioning consultant.
TASK:
Sharpen our current ICP based on the data below.
CURRENT ICP:
[PASTE OR DESCRIBE]
RECENT CUSTOMERS (last 10-20 wins):
[PASTE LIST WITH COMPANY, SIZE, INDUSTRY, USE CASE]
LOSSES (last 10):
[PASTE LIST WITH REASON]
CONSTRAINTS:
- Find patterns in wins that the current ICP doesn't capture.
- Flag segments where we consistently lose — and why.
- Output an ICP sharp enough that a BDR could use it to qualify inbound in 30 seconds.
OUTPUT:
- Refined ICP (firmographics + trigger events)
- Who we should stop targeting (with reasoning)
- 3 qualifying questions a BDR can ask on a first call
Swap: [CURRENT ICP], [RECENT WINS], [RECENT LOSSES].
Tip: The output improves dramatically when you feed actual win/loss notes rather than just industry/size data.
27. JTBD interview planner
What it does: Designs a Jobs-to-be-Done interview script for customer discovery.
Prompt:
You are a JTBD-trained product marketer.
TASK:
Design an interview script for customer discovery using the Jobs-to-be-Done framework.
PRODUCT:
[PRODUCT]
GOAL OF RESEARCH:
[e.g. "understand why users switch TO us from competitor X"]
INTERVIEWEE:
[RECENT CUSTOMER / CHURNED USER / PROSPECT]
CONSTRAINTS:
- 45-minute interview.
- Open-ended questions, no leading prompts.
- Must surface the trigger moment (when they decided they needed a solution).
- Must surface the struggle (what they tried before).
OUTPUT:
- Warm-up questions (5 min)
- Trigger discovery (10 min)
- Struggle discovery (15 min)
- Evaluation process (10 min)
- Usage + outcome (5 min)
- Closing question (open-ended)
- Notes for the interviewer on what to listen for
Swap: [PRODUCT], [GOAL OF RESEARCH], [INTERVIEWEE].
Tip: Run 5 interviews, then feed all 5 transcripts into Claude for pattern-matching with the positioning prompt (#29).
28. Competitor teardown
What it does: Breaks down a competitor's marketing strategy and finds weaknesses.
Prompt:
You are a competitive intelligence analyst.
TASK:
Do a strategic teardown of the competitor below.
COMPETITOR:
- Name: [COMPETITOR]
- Homepage: [URL]
- Their key messaging (paste 2-3 hero headlines): [PASTE]
OUR PRODUCT:
[ONE PARAGRAPH]
CONSTRAINTS:
- Do not guess unknowable things (revenue, team size) unless we've given the data.
- Analyze: positioning, pricing angle, target persona, core promise.
- Find 3 gaps where we can clearly differentiate.
OUTPUT:
- Competitor positioning (in their own words, paraphrased)
- Their implied ICP
- Their core promise
- 3 weaknesses or gaps
- 3 positioning angles we should own that they don't
Swap: [COMPETITOR], [URL], [HERO HEADLINES], [OUR PRODUCT].
Tip: Compare Claude's output across 3 competitors in one go to see which gaps are universal vs. competitor-specific.
29. Positioning doc draft
What it does: Produces a first-draft positioning doc using April Dunford's structure.
Prompt:
You are a positioning strategist trained on April Dunford's framework.
TASK:
Write a first-draft positioning doc for our product.
PRODUCT:
[PRODUCT DESCRIPTION]
WHO CURRENTLY BUYS US:
[2-3 WIN PROFILES]
WHAT WE REPLACE:
[CURRENT SOLUTIONS THEY USED BEFORE]
OUR UNIQUE CAPABILITIES:
[LIST 3-5]
CONSTRAINTS:
- Use April Dunford's 5 components: competitive alternatives, unique attributes, value, who cares, market category.
- No buzzwords — if I can't explain it to my mom, rewrite it.
- Flag anything you had to guess vs. what we gave you.
OUTPUT:
- Competitive alternatives (what do they use today)
- Unique attributes (what only we have)
- Value (the benefit those attributes unlock)
- Who cares a lot (which customer segment cares most)
- Market category we position in
- Homepage H1 and subhead that express this positioning
Swap: [PRODUCT], [WIN PROFILES], [CURRENT SOLUTIONS], [UNIQUE CAPABILITIES].
Tip: Use this as input to a real positioning workshop, not as the final output. It gives teams a concrete starting point to react to — which beats a blank page every time.
30. Messaging hierarchy
What it does: Maps your positioning into a messaging hierarchy for every channel.
Prompt:
You are a brand strategist.
TASK:
Build a messaging hierarchy from the positioning below.
POSITIONING:
[PASTE POSITIONING DOC OR SUMMARY]
CONSTRAINTS:
- One primary message (headline everything cascades from).
- Three supporting pillars.
- Under each pillar: 3 proof points + 1 tagline.
- Must adapt to homepage, ads, email, and sales decks.
OUTPUT:
- Primary message
- 3 pillars with their proof points and taglines
- Primary message translated for: homepage hero, paid ad, cold email subject, sales deck opener
Swap: [POSITIONING].
Tip: Save this hierarchy as a Claude Project. Every new campaign starts by pulling the right pillar instead of reinventing the wheel.
Pro tips for marketing with Claude
Five things that separate marketers getting 5x output from Claude vs. those who type one-liners and give up:
- Use Claude Projects for brand context. Drop your positioning doc, tone guide, top 3 best-performing emails, and ICP into a Claude Project's instructions. Every future prompt inherits that context — you stop re-pasting it every time.
- Pick the right Claude model. Claude Opus 4.7 is worth the cost for strategy, positioning, and long-form writing. Sonnet 4.6 is better for high-volume execution (50 ad variants, 30 emails). Haiku for simple classification tasks.
- Give Claude examples of your best work. Two paragraphs of your highest-performing past copy beats 500 words of brand guidelines.
- Specify the format, not just the content. "Output as a table with columns X, Y, Z" produces usable work. "Write about our campaign" produces a blog post you'll have to reformat.
- Use Artifacts for deliverables. Claude's Artifacts feature turns a prompt into a saveable, shareable deliverable — particularly useful for briefs, one-pagers, and SQL result summaries.
For more on Claude prompting patterns that work for marketing teams, see Claude prompt engineering best practices. For a deeper look at how Claude compares to ChatGPT on marketing tasks, see Claude vs. ChatGPT.
FAQ
What's the best Claude model for marketing prompts?
Claude Opus 4.7 is the strongest for strategy, positioning, and long-form content. Claude Sonnet 4.6 is the best price/performance ratio for high-volume execution like ad copy variants or email drafts. Haiku is best for fast, repetitive classification tasks (tagging leads, categorizing feedback). For most marketers, Sonnet 4.6 is the default, with Opus reserved for high-stakes strategy work.
Can I use these Claude marketing prompts with Claude Free?
Yes. Every prompt in this post works on the free tier, which uses Claude Sonnet. You'll hit usage limits faster than paid plans, and you won't get Opus access, but quality stays high for everyday marketing work.
How do I customize these for my brand voice?
Paste any prompt into the Claude prompt generator, add a section describing your brand voice in 3 adjectives, and include 2–3 sentences of example copy that represents your voice. Claude matches voice better from examples than from rules. Even better: set this up once as a Claude Project's custom instructions, and every future prompt inherits it.
Where can I save these Claude prompts for reuse?
Three options, in order of power: (1) Claude Projects — save each prompt with brand context in the project instructions; (2) a Notion database with one prompt per row; (3) Prompt Builder's account-saved prompts — sign up free and save any prompt you've generated.
What marketing tasks shouldn't I use Claude for?
Three: (1) anything that requires real-time data (prices, stock, competitor launches this week) — Claude doesn't browse by default; (2) final legal copy — always have a lawyer review; (3) prompts that require personal data about individuals — keep PII out of prompts unless you're on an enterprise plan with a BAA.
Ready to customize any of these for your exact brand and campaign?
→ Generate a custom Claude marketing prompt in 10 seconds
For the full hub of Claude marketing prompts organized by use case, see /prompts/marketing/claude.
