10 Claude Prompts for Marketing That Converts
Marketing is where AI output quality varies most wildly — because marketing prompts are usually missing the two things marketing actually runs on: a sharply defined customer and a reason to believe. Give Claude those, and it stops producing brochure-speak.
These ten prompts are built around real marketing craft: positioning before copy, customer language before cleverness, one job per asset. They work for SaaS, e-commerce, services, and personal brands — swap the placeholders and go.
How to use these prompts
- Paste real customer quotes wherever you have them — reviews, support tickets, sales-call notes. Customer language outperforms invented copy in almost every test.
- Define ONE reader per asset. Copy written for 'businesses of all sizes' converts none of them.
- Give Claude your actual differentiator and the proof for it. Without proof, it will generate claims — and claims without proof read as noise.
- Generate variants, then test. Claude's job is a strong starting field, not a guaranteed winner.
The prompts
1. The positioning sharpener
Sharpen my positioning using April Dunford's framework. Product: [what it is and does] Customers who love us most: [who they are, and why they chose us — real quotes if you have them] Alternatives they compare us to: [including "do nothing" and spreadsheets] What we do that alternatives can't: [features/capabilities — be concrete] Work through: 1. True competitive alternatives (from the customer's view, not our category's). 2. Our differentiated capabilities — and cut any that customers don't actually care about. 3. The value those capabilities enable, in customer outcomes. 4. Who cares most about that value (the segment where we win by default). 5. The market category that makes all this obvious in one sentence. Then write the positioning statement and the 3 messages everything else should ladder up to.
2. The landing page architect
Write a landing page for [product/offer]. The ONE reader: [specific person — role, pain, what they were doing 5 minutes before landing here] Traffic source: [where they come from — this sets what they already know] The action: [the single conversion goal] Proof we have: [numbers, testimonials, logos, guarantees — only what's real] Structure: 1. Hero: headline (their outcome, their words), subhead (how + for whom), CTA. 2. Problem section that shows we understand their current painful workaround. 3. How it works — 3 steps max. 4. Proof section using ONLY the proof I gave you, placed for maximum effect. 5. Objection handling as FAQ: the 4 real reasons this reader wouldn't buy. 6. Final CTA with risk reversal. Write actual copy, not descriptions of copy. Match sophistication to the traffic source.
3. The customer-language miner
Mine this raw customer language for marketing gold: [paste reviews / support tickets / sales call notes / survey answers / Reddit threads] Extract and organize: 1. Exact phrases customers use for the problem (their words — flag anything we'd never have written ourselves). 2. The "moment of struggle" stories: what happened right before they went looking for a solution? 3. Objections and anxieties, in their words. 4. The outcome language: how do they describe success after buying? 5. Turn the 5 strongest phrases into: 2 headlines, 2 email subject lines, 1 ad hook — using their wording nearly verbatim.
4. The email sequence builder
Build a [5]-email sequence for: [goal — e.g. "trial users who haven't activated" / "post-purchase onboarding" / "cold leads from the ebook"] Reader: [who, and their state of mind entering the sequence] Product: [what, and the ONE action each email should drive toward] Voice: [paste a sample email you like] For each email: - Trigger/timing and the job of this specific email in the sequence - Subject line (2 options: one curiosity, one direct) - Body under 150 words, one idea, one CTA - The exit condition (when someone should stop getting the sequence) Sequence logic: each email must earn the next — no "just checking in" filler. Email 3 should handle the biggest objection head-on.
5. The SEO content brief
Create a content brief for an article targeting: [keyword/topic] Searcher intent: [what is this person actually trying to do when they search this?] Our angle: [what we can say that the current top results can't — experience, data, contrarian take] Top-ranking content right now: [paste titles/summaries of the top 3–5 results] Brief: 1. The intent gap: what do current results fail to answer? That's our headline advantage. 2. Title (3 options) and H2 outline — ordered by searcher priority, not our narrative preference. 3. Questions to answer for featured snippets / AI answer engines (with the direct-answer sentence for each). 4. What genuinely unique thing must this article contain to deserve to rank? If we have nothing, say so — that's a "don't write it" verdict. 5. Internal links it should give and receive.
6. The ad variant machine
Generate ad variants for [platform: Meta / Google / LinkedIn / X].
Offer: [what's being promoted]
Audience: [targeting — who sees this]
Proof: [the strongest specific: number, testimonial, outcome]
Winning ads so far: [paste any current winners — Claude should learn from them, not repeat them]
Give me 10 variants across distinct angles:
- 2 pain-first (the moment of struggle)
- 2 proof-first (lead with the number)
- 2 identity ("for people who...")
- 2 objection-flip (attack the reason they scroll past)
- 2 curiosity (specific, not clickbait)
For each: hook line, body (platform-appropriate length), CTA. Then predict the winner and say why — I want your reasoning, not just output.7. The launch announcement stack
We're launching [thing]. Write the full announcement stack, one core message, every format: The launch: [what's new, who it's for, why it matters] The ONE takeaway everyone should remember: [single sentence] Proof/demo: [what we can show] Write: 1. X/Twitter thread (6–8 posts, first post must work standalone) 2. LinkedIn post (professional but human, no "I'm thrilled to announce") 3. Email to existing users (angle: what this makes possible for them, not what we did) 4. Product Hunt tagline + first-comment maker story 5. 2-sentence version for a newsletter mention Same message, native to each format — don't just resize the same paragraph. Ban: "excited", "thrilled", "game-changer", "revolutionize", rocket emoji.
8. The pricing page psychologist
Review my pricing page like a conversion specialist: [paste pricing page copy/structure, or describe tiers and prices] Customer context: [who buys, deal size, sales-assisted or self-serve] Analyze: 1. Which tier is doing the anchoring, and is it anchoring in the right direction? 2. Tier naming and feature-gating: does the middle tier make the decision easy? (It should be ~70% of buyers.) 3. Where does cognitive load spike — too many rows, unclear units, "contact us" ambiguity? 4. The FUD audit: what unanswered anxiety sits next to the buy button (lock-in, migration, what happens at the limit)? 5. Three specific changes ranked by expected impact, and the one A/B test to run first.
9. The case study interviewer & writer
Turn this customer win into a case study. Raw material: [interview notes / email thread / metrics — whatever exists] Customer: [who, industry, size — and what I'm allowed to name publicly] Reader: [the prospect persona this case study should convince] First: list the 5 follow-up questions I should ask the customer to make this 10x stronger (missing numbers, the before-state, the moment they knew it worked). Then write it with what we have: - Title: their outcome with a number, not our product name - Before: the struggle in their words - Decision: why us over alternatives (honest, including hesitations) - After: outcomes with numbers; the workflow that changed - Pull quote that a salesperson would actually use Keep it under 600 words. Mark every place where a real number would strengthen a vague claim.
10. The funnel diagnostician
Diagnose my funnel and tell me what to fix first. Funnel stages & numbers: [e.g. "10k visitors → 400 signups (4%) → 120 activated (30%) → 25 paid (21%)"] Benchmarks I know of: [if any] What we've already tried: [so Claude doesn't suggest it again] 1. Which conversion is most below reasonable benchmarks for [business type]? Show comparative reasoning. 2. For the worst stage: the 3 most likely causes, and the fastest way to distinguish between them (session recordings? exit survey? one interview question?). 3. The math: if we fix the worst stage to benchmark, what happens to revenue vs. fixing any other stage? (Leverage, not vibes.) 4. This week's action: one diagnostic, one experiment, expected learning from each.
Getting more out of every prompt
Customer words beat clever words
The mining prompt isn't one of ten — it's the one to run first. Every other prompt gets better when you feed it real voice-of-customer language.
One asset, one reader, one job
Every conversion asset in this list demands a single reader and single action. When Claude's marketing output feels mushy, it's almost always because the prompt asked it to serve three audiences at once.
Proof is your bottleneck, not copy
Claude can generate infinite claims. Only you can supply the numbers, testimonials, and demos that make claims land. Gather proof first; prompt second.
Ship variants, keep score
Treat outputs as a test field. Track which angles win in your market, tell Claude the winners next time, and it compounds — the ad-variant prompt explicitly builds this loop.
Want this expertise installed, not pasted? The ClaudeThings kits turn workflows like these into 89 agents, 103 skills, and 181 slash commands that live inside Claude Code — one command to install, any stack. See the kits →