Forty Claude prompts for real business work, eight each for marketing, sales, product management, operations and HR. All of them use the same four-layer structure: a role, the context, the exact task, and the output format. Copy them as they are, or read the note under the ones where a single line is doing most of the work.
The four layers, before you copy anything
Every prompt below is built the same way, and knowing the pattern matters more than the forty examples, because it lets you write the forty-first.
- Role. Not decoration. It selects the register the answer gets written in.
- Context. Product, audience, and the constraints that make your case yours. Anything in square brackets is a slot you fill.
- Task. The deliverable, named. "Five discovery questions" not "some questions".
- Format. Structure and length. Skip it and you get prose where you wanted a table.
The fifth thing, optional but powerful: say what to leave out. Several prompts below ban a specific phrase, and that single line does more for the tone than any instruction to sound human.
Marketing (8 prompts)
1. Brand voice audit
Act as a brand strategist. Review these three pieces of our content: [PASTE]. Identify the implied brand voice, flag every inconsistency across the three, and give me three specific rules the team should follow to make it consistent. Rules must be checkable, not aspirational.
The last sentence is the one that matters. Without it you get "be authentic"; with it you get "never open with a question".
2. Landing page copy
Write landing page copy for [PRODUCT] targeting [AUDIENCE]. Include: hero headline and subheadline, three benefit sections each with a heading and a two-sentence description, a social proof placeholder telling me what kind of proof to go and collect, an FAQ with four questions, and a closing CTA section. Brand tone: [TONE]. Do not use the words innovative, seamless or empower.
3. Monthly content calendar
Build a four-week content calendar for [BRAND] in [INDUSTRY]. Three posts a week across LinkedIn, Instagram and email. For each: platform, format, topic, hook line, and goal (awareness, engagement or conversion). Keep the mix 70 percent value-led, 30 percent commercial. Flag any week where the same idea appears twice in different clothes.
4. Competitive positioning map
Here are four competitors: [LIST]. For each, give their positioning angle, target customer, pricing tier and the evidence for it, and one gap they are not addressing. Then propose a positioning statement for us that occupies the clearest white space. If two of them are effectively saying the same thing, say so first.
Convergent competitor language is the most reliable positioning gap there is, and you cannot see it reading the sites one at a time.
5. Ad copy testing pack
Write a testing pack for a Facebook campaign promoting [OFFER]. Four headline variants under 40 characters, each testing a different angle: fear, aspiration, social proof, curiosity. Four primary text variants under 125 characters, one matched to each headline. Label the psychological mechanism in each. Do not repeat a mechanism across variants.
6. Press release draft
Write a press release for [ANNOUNCEMENT]. Headline, dateline, opening paragraph covering who, what, when, where and why in two sentences, two supporting paragraphs, one executive quote with the name left blank, boilerplate about [COMPANY], and a media contact placeholder. AP style, under 500 words. The quote must say something a competitor could not also say.
7. SEO blog outline
Create a blog outline targeting the keyword [KEYWORD]. Include: H1, meta description under 155 characters, six H2 sections with a one-sentence description each, three internal linking opportunities, and a suggested word count based on what the query seems to want. The post must answer the search intent inside the first 100 words. Tell me which of the six sections is the one competitors are likely to skip.
Want this run end to end rather than prompt by prompt? See AI SEO Agents.
8. Product launch checklist
Build a launch checklist for a [PRODUCT TYPE] going to [AUDIENCE] in [TIMEFRAME]. Organise by phase: six weeks out, launch week, two weeks after. Five to eight tasks per phase with owner role, priority and dependencies. Output as a table. Mark any task that, if it slips, blocks the launch date.

The eight prompts above each produce one asset. Luce designs the journey they hang off: channels, content, conversion events and the metrics that tell you which stage is leaking.
View Luce →Sales (8 prompts)
The complete set lives in the Sales Prompts collection.
9. ICP definition
Help me define our Ideal Customer Profile. Product: [PRODUCT]. Our best current customers: [DESCRIBE TWO OR THREE]. Build an ICP card covering company size, industry, tech stack signals, the trigger events that make them ready to buy, the role who champions the purchase, and the economic buyer. Be specific, no ranges wider than 50 percent. Then list what our best customers have in common that is NOT in the card.
10. Five-part cold sequence
Write a five-email cold sequence to [ROLE] at [COMPANY TYPE]. Product: [PRODUCT]. Pain we solve: [PAIN]. Email 1 value-first with no ask. Email 2 social proof. Email 3 a different value frame. Email 4 a soft ask for 15 minutes. Email 5 a breakup that leaves the door open. Max 100 words each, subject lines included. No email may open with the word I.
That last constraint is a cheap trick that works: it forces every opening line to be about the reader.
11. Demo script outline
Build a 30-minute demo script for [PRODUCT] to [ROLE] at [COMPANY TYPE]. Five minutes of discovery to open, twenty minutes walking through three core flows, five minutes to close on next steps. For each section: talking points, questions to ask, and the red flags to listen for. Include a slide-by-slide outline and mark the one moment where the deal is usually won or lost.
12. Win/loss analysis
I lost a deal to [COMPETITOR]. Here is what I have: [PASTE NOTES, EMAILS OR CRM RECORD]. Give me: the stated reason, the likely real reason based on the pattern, the signals I missed and where in the process they appeared, and one thing I could have done differently at each stage. Be direct. I want diagnosis, not sympathy.
13. Discovery question set
Write twelve discovery questions for a first call with [ROLE] at [COMPANY TYPE] about [PROBLEM]. None answerable with yes or no. Order them so the easy context questions come first and the budget and authority questions come last. For each, add the follow-up probe to use if the answer is vague.
14. Pricing objection playbook
We sell [PRODUCT] at [PRICE] to [ICP]. Write a playbook for the five most likely pricing objections. For each: the objection in the buyer's own words, what it usually means underneath, a response under 80 words, and the one question to ask before responding at all. Do not suggest discounting in any of them.
15. Account research brief
Here is what I know about [COMPANY]: [PASTE]. Build a one-page pre-call brief: what they do in one sentence, three recent changes that could create urgency, who likely owns this problem, two things they are probably already trying, and the single question that would tell me fastest whether this is real. Flag anything you are inferring rather than reading.
16. Follow-up after silence
A prospect went quiet after [WHAT HAPPENED]. Last contact: [DATE AND CONTEXT]. Write three follow-ups of increasing directness, each under 60 words. One re-opens with new value, one names the silence plainly, one closes the file. Do not use just checking in, circling back, or bumping this to the top of your inbox.

Builds a structured discovery, demo or close script in about ten minutes. The ready-made companion to prompts 11 to 13 above.
View Sales Script Prompt →Product management (8 prompts)
17. PRD first draft
Write the first draft of a PRD for [FEATURE]. Include: problem statement covering who has it, why it matters and what they do today instead; proposed solution; at least five user stories in standard format; success metrics; explicit out-of-scope items; and open questions. Structured document with headers. The out-of-scope section must have at least three items.
Forcing out-of-scope items is the trick. A PRD without them is a wish list, and the argument you avoid now happens in week three of the build.
18. Roadmap prioritisation
Here is our backlog: [PASTE LIST]. Score each item with RICE. Where you do not have enough information to score confidently, ask me rather than estimating. Output a ranked table with the reasoning for the top three and the bottom three. Then tell me which item the ranking is most sensitive to, so I know which estimate to sanity-check.
19. User interview question set
I am interviewing [WHO] about [PRODUCT OR FEATURE]. What I need to learn: [GOAL]. Write twelve questions: warm-up context first, then exploratory behaviour, then deep on pain and current workarounds. Three follow-up probes per key question. Remove any question that invites the person to predict their own future behaviour.
That last instruction removes the questions that generate the most confident and least reliable research data.
20. Feature spec review
Here is a feature spec: [PASTE]. Act as a sceptical senior engineer. List: the assumptions that would break the estimate if wrong, the edge cases the spec does not mention, the dependencies on other teams, and the one part most likely to take three times longer than planned. Do not suggest improvements, only risks.
21. Release notes from a changelog
Here is our raw changelog: [PASTE]. Write release notes for [AUDIENCE]. Group by what changed for the user, not by component. Lead with the change most people will notice. Skip anything invisible to a user unless it fixes something they complained about. Plain language, no version numbers in the headings.
22. Metric definition sheet
We want to track [METRIC NAME] for [PRODUCT]. Write a definition sheet: the precise formula, what counts and what does not, the edge cases that make it ambiguous, how it can be gamed, and one companion metric that would catch the gaming. Say plainly if the metric is a poor proxy for what we actually care about.
23. Competitor feature teardown
Here is a competitor's feature: [DESCRIBE OR PASTE DOCS]. Work out what problem they were solving and for whom, what they deliberately left out, what it implies about their roadmap, and whether copying it would help or hurt our positioning. Answer the last part with a recommendation, not options.
24. Stakeholder update
Write a fortnightly product update for [AUDIENCE: exec team, wider company, board]. Sections: what shipped, what moved, what is blocked and what I need from you. Under 300 words. Lead with the thing this audience actually cares about, and tell me which one that is before you write it.
Operations (8 prompts)
Developer-focused prompts are in Tech and Development Prompts.
25. Process documentation
I will describe a process we run manually. Turn it into an SOP: process name, owner, frequency, trigger, numbered steps with one action each, decision points in if-then form, escalation rules and quality checks. Flag any step where I have described the what but not the how. Here is the process: [DESCRIBE].
26. Vendor evaluation scorecard
Build a vendor scorecard for choosing a [VENDOR TYPE]. Eight criteria relevant to this category, a 1 to 5 scoring guide with a concrete description of what a 3 looks like, and a weighting system. Output as a table I can fill in per vendor. Include at least one criterion that vendors in this category usually hope buyers forget to ask about.
27. Incident post-mortem
Here is what happened: [DESCRIBE INCIDENT AND TIMELINE]. Write a blameless post-mortem: timeline, impact in customer terms, contributing causes, what made detection slow, what made recovery slow, and action items with an owner role and a type (prevent, detect or mitigate). No action item may be train the team better.
28. Capacity plan
We are a team of [SIZE] doing [WORK TYPE] with a volume of [VOLUME]. Build a capacity plan: current utilisation, the point at which quality starts to drop, the leading indicator that tells us we are past it, and three options at different costs. State the assumption each option depends on.
29. Tool consolidation audit
Here is our current tool stack with costs: [PASTE LIST]. Identify overlaps in function, tools likely used by fewer than three people, contract timing risks, and the switching cost that makes an obvious consolidation not worth it. Rank by annual saving against migration pain, and name the one you would not touch.
30. Meeting agenda that earns the hour
We meet [FREQUENCY] for [DURATION] about [TOPIC], attendees [ROLES]. Write an agenda where every item has a decision to be made or an input needed, with a time box and the person who owns it. Move anything that is just information into a written pre-read. If the remaining agenda does not fill the slot, say so and recommend a shorter meeting.
31. Handover document
I am handing [ROLE OR PROJECT] to someone else. Here is what I do: [DESCRIBE]. Write a handover: recurring commitments with timing, the systems and access needed, the relationships that matter and why, the decisions currently in flight, and the undocumented things that would take a new person months to work out. Ask me about anything you can tell is missing.
32. Policy written for humans
Draft a [POLICY TYPE] policy for a company of [SIZE] in [COUNTRY]. Plain language, no legal boilerplate. Cover what it applies to, what people should do, who decides the edge cases, and what happens if it is not followed. Under 500 words. Mark the parts that need a lawyer's eyes before publication.
HR and people (8 prompts)
33. Job description builder
Write a job description for a [TITLE] at a [COMPANY TYPE], seniority [LEVEL]. The role will: [THREE RESPONSIBILITIES]. Must-haves: [LIST]. Nice to have: [LIST]. Direct and honest, no fast-paced environment and no passionate team player. Under 400 words. Include a section on what makes this role different from the same title elsewhere.
34. Performance review framework
Build a review framework for [ROLE]. Five evaluation dimensions specific to this job, not communication and teamwork. A four-point scale with a behavioural anchor describing what each level actually looks like. Three development questions. A feedback structure that pushes the manager toward examples instead of adjectives.
35. Structured interview scorecard
Build an interview scorecard for [ROLE]. Four competencies drawn from what the job actually requires, two behavioural questions per competency, and what a strong versus weak answer sounds like for each. Add the one question that best separates people who have done this job from people who have read about it.
36. Thirty-sixty-ninety onboarding plan
Write a 30-60-90 plan for a new [ROLE] joining [TEAM]. For each phase: what they should understand, what they should be doing independently, who they should have met and why, and the checkpoint that tells us it is going well. Make day one and week one concrete down to the hour.
37. Compensation conversation prep
An employee in [ROLE] has asked for a raise. Context: [PERFORMANCE, TENURE, CURRENT BAND, BUDGET REALITY]. Prepare me: what to establish before answering, how to say yes, how to say not now without losing them, and the three things not to say. Assume they have another offer even if they have not mentioned one.
38. Engagement survey question set
Write fifteen engagement survey questions for a company of [SIZE]. Mix of scale and open questions. Each one must map to something we could actually change. Remove any question we would not act on, and tell me which ones will produce the most uncomfortable answers so we can decide now whether we mean it.
39. Difficult conversation script
I need to raise [ISSUE] with [ROLE]. Background: [CONTEXT]. Write an opening under 60 words that names the issue without a compliment sandwich, three likely responses and how to handle each, the outcome I should be aiming for, and the line I should not cross. Tell me if this should be a conversation or a written warning.
40. Exit interview question set
Write twelve exit interview questions for a [ROLE] leaving after [TENURE]. Order them so the easy ones build enough trust for the honest ones. Include questions that surface problems they may not blame on us. Avoid anything that invites a polite non-answer, and mark the two questions worth asking again six months later.

Runs evidence-based reviews, interview scorecards and the difficult conversations around them, without rebuilding the structure from scratch each time.
View Claire →When to stop pasting prompts
These forty share one limit: every new chat starts from nothing. You paste the prompt, refill the context, get the output, close the tab, and do it again on Thursday.
A skill file is where that stops. It is a ZIP with a SKILL.md inside, uploaded once in the Claude desktop app under Customize → Skills, and Claude loads it whenever your request matches what it is for. On a free plan, or in ChatGPT and Gemini, paste the same file into custom instructions. Step by step: how to install a Claude skill.
Prompt packs are $9 and give you more of what is on this page. Role skills are $14.99 and give you the method behind it. If you have filled in the same square brackets three weeks running, buy the skill.
In summary:
Role, context, task, format, plus one thing to leave out. For marketing, the AI Marketing Funnel Builder ($9) ties the first eight into one system; for sales, the Sales Script Prompt ($9) turns discovery, demo and close into a repeatable script; for HR, the Performance Review Prompt ($9) packages prompt 34. All work in Claude, ChatGPT and any AI chat, no coding, 30-day money-back guarantee.
Skip the copy-paste. Organised by function, ready to run in Claude, ChatGPT or any AI chat.
Browse business prompts →Browse all prompts →Related reading
- Claude skills by role - find the one for your exact job.
- AI business skills for Claude - operations, strategy, project management, reporting.
- AI agents for business - the complete guide for 2026.
- The Claude skills marketplace - what you are actually buying.