A Claude Prompt Library for Marketers: 50 Prompts to Adapt

#01How do you use this library?

Every prompt follows the pattern from lesson 3: say what you want, give the context Claude can't know, describe the output, and set constraints. A few habits make them work far better.

Work inside a project. Run them in the Claude project that holds your brand voice, audience notes and fact sheet. Then "our audience" and "our product" mean something.

Fill every bracket. "[audience]" should become "finance managers at 50 to 500 person companies who approve invoices weekly", not "finance people".

Ask for questions first on big jobs. Adding "Before you start, ask me up to five questions about anything you'd need to do this well" to any prompt below usually improves the result more than any wording change.

Iterate in the same chat. The first output is a draft. Say what's wrong specifically ("the second paragraph promises too much; our product doesn't do X") rather than starting again.

Our view: the prompt matters less than the context. A plain prompt with a good brief, real customer quotes and your fact sheet beats a clever prompt with none of them.

#02Strategy and positioning prompts

1. POSITIONING CHECK
Here is our positioning statement: [paste]. Here are three
competitor homepages: [paste text]. Where does our positioning
overlap with theirs, and what could we credibly claim that
none of them do? Use only what's in the text I gave you.

2. MESSAGE HOUSE
Build a message house for [product] aimed at [audience]: one
umbrella message, three pillars, and two proof points per
pillar. Use only proof points from the fact sheet. Mark any
pillar that lacks proof as [NEEDS PROOF].

3. OBJECTION MAP
List the ten objections a [role] is most likely to raise before
buying [product category]. For each, say which stage it
appears at and what evidence would answer it. Then tell me
which ones our current site doesn't address: [paste site copy].

4. LAUNCH TIERING
We're launching [feature]. Here's what it does and who it's
for: [details]. Recommend a launch tier (major, medium, minor)
with reasons, and list the channels and assets each tier would
need. Flag any assumption you've made.

5. PERSONA FROM DATA
From these interview notes and reviews [paste], draft a
persona for [segment]: jobs to be done, triggers, objections,
words they use. Quote the source for every point. Don't add
anything you can't trace to the material.

#03Research prompts

6. COMPETITOR TEARDOWN
Search the web for [competitor]'s current pricing, homepage
claims and recent announcements. Summarise in a table with a
source link for each row. Say clearly where you couldn't find
something.

7. SEARCH INTENT
For the query "[query]", search the web and describe what the
top results have in common: format, angle, depth, and what
questions they answer. Then tell me what a better page would
need to include.

8. REVIEW MINING
Here are [n] customer reviews: [paste]. Group them into themes,
count how many mention each, and give two verbatim quotes per
theme. Separate praise from complaints.

9. MARKET QUESTION LIST
List the 30 questions a [role] would ask an AI assistant when
choosing a [category] tool, grouped by stage: problem,
options, comparison, decision. Use their language, not ours.

10. TREND CHECK
Search for news and announcements about [topic] from the last
[period]. Give me the five developments that matter most for
[audience], each with a source and one sentence on why it
matters. Say if you found fewer than five.

#04Content creation prompts

11. BRIEF FROM A TOPIC
Write a content brief for "[topic]" for [audience]: search
intent, the one question it must answer, outline with H2s
written as questions, proof points from the fact sheet, and
internal links from this list: [paste URLs].

12. DRAFT FROM BRIEF
Write the article from this brief: [paste]. Answer the main
question in the first two sentences. Use short paragraphs.
Where you need a fact that isn't in the brief or fact sheet,
write [CHECK: what's needed].

13. INTRO OPTIONS
Give me five different openings for this piece [paste], each
under 50 words, each taking a different angle: a question, a
number from our fact sheet, a customer quote, a contrarian
view, a direct answer.

14. FAQ BLOCK
From this page [paste], write six FAQs a [role] would actually
ask, with answers of two to three sentences that stand on
their own. Use only information on the page.

15. CASE STUDY DRAFT
Turn this interview transcript [paste] into a case study:
challenge, why they chose us, how they use it, results. Use
their words in quotes. Only include results they actually
stated, with the exact figures.

#05Editing and quality prompts

16. FACT CHECK LIST
List every factual claim in this draft [paste]. For each, say
whether it's supported by the fact sheet, by a source in the
draft, or by nothing. Don't rewrite anything.

17. CUT BY A THIRD
Shorten this to two thirds of its length [paste] without
losing any fact or example. Show me what you removed as a
list at the end.

18. VOICE CHECK
Compare this draft [paste] with our brand voice guidelines.
List the five sentences that sound least like us and suggest
a rewrite for each. Keep everything else unchanged.

19. READER TEST
Read this as a sceptical [role] who has never heard of us.
What's unclear, what's unbelievable and what's missing? Be
blunt and specific.

20. HEADLINE SHORTLIST
Write ten headlines for this piece [paste], under 60
characters each. Then pick the three that best match the
search intent "[query]" and explain why.

#06Repurposing prompts

21. ONE TO MANY
Turn this article [paste] into: one LinkedIn post, three X
posts, one newsletter section, a 60-second video script and
five quote cards. Each must stand on its own and link back.

22. WEBINAR TO BLOG
Here's a webinar transcript [paste]. Write a blog post that
covers the three most useful points, with the speaker's best
quotes, and a short section answering audience questions.

23. LONG TO SHORT EMAIL
Turn this article [paste] into a 150-word email for [segment]
with one clear call to action: [CTA].

24. SLIDE OUTLINE
Turn this report [paste] into a ten-slide outline: a headline
that states the point on each slide, three supporting
bullets, and the chart or visual each slide needs.

25. SALES ENABLEMENT
Turn this launch page [paste] into a one-page sales brief:
what's new, who it's for, three talking points, likely
objections and answers, and what not to promise.

#07SEO and AI visibility prompts

26. TITLE AND META
Write five title tags (under 60 characters) and five meta
descriptions (under 155 characters) for this page [paste],
targeting "[query]". Don't use clickbait.

27. CONTENT GAP
Here's our page [paste] and the top three ranking pages for
"[query]" [paste]. What do they cover that we don't? List
gaps in order of importance to the reader.

28. ANSWER-FIRST REWRITE
Rewrite the first paragraph of each section in this page
[paste] so it directly answers the section heading in one or
two sentences. Leave the rest unchanged.

29. ENTITY CHECK
Read our About page and product pages [paste]. List
inconsistencies in how we describe our company, product names,
categories, pricing or founding facts. AI tools get confused
by these.

30. CITATION READINESS
Review this page [paste] as if you were deciding whether to
cite it in an answer about "[question]". What makes it easy
or hard to quote? Suggest specific changes.

#08Paid media prompts

31. ANGLE GENERATOR
Give me ten distinct ad angles for [product] aimed at [role],
each based on a different motivation (save time, reduce risk,
look good to the boss, and so on). One sentence per angle.

32. SEARCH AD ASSETS
Write 15 headlines (30 characters max) and 4 descriptions
(90 characters max) for [product] targeting "[keyword]".
Count characters and show the count next to each.

33. AD VARIANTS FOR A TEST
Take this winning ad [paste] and write four variants that
each change only one thing: the hook, the proof, the offer
or the call to action. Label what changed.

34. LANDING PAGE MATCH
Here's the ad [paste] and the landing page [paste]. Does the
page deliver on the ad's promise in the first screen? List
mismatches and suggest fixes.

35. RESULTS READOUT
Here's a CSV export of ad results [upload]. Which ads and
audiences perform best on [metric]? Show the numbers, say
how confident we can be given the volumes, and flag anything
too small to judge.

#09Email and lifecycle prompts

36. ONBOARDING SEQUENCE
Plan a five-email onboarding sequence for new [product]
users. For each: goal, trigger, subject line, one-sentence
summary and the single action we want.

37. SUBJECT LINES
Write ten subject lines for this email [paste], under 50
characters, in different styles. Flag any that could trigger
spam filters or overpromise.

38. WIN-BACK
Write a win-back email for customers who stopped using
[feature] for [period]. Be honest about why they might have
left. Offer help, not pressure.

39. PERSONALISATION SAFELY
Here are our segments [list, no personal data]. Write one
version of this email [paste] per segment, changing only the
opening and the example used.

40. NEWSLETTER EDIT
Here are this month's items [paste]. Write a newsletter with
a short intro, the three items that matter most first, and
one sentence on why each matters to [audience].

#10Social prompts

41. CONTENT CALENDAR
Plan four weeks of LinkedIn posts for [brand] from these
themes [list] and assets [list]. Two posts a week. For each:
date, format, hook and the asset it uses.

42. POST FROM AN INSIGHT
Turn this insight [paste] into a LinkedIn post: a first line
that stops the scroll without clickbait, three short
paragraphs, and a question at the end. No hashtags in the
body.

43. THREAD
Turn this article [paste] into an X thread of six to eight
posts. The first post must make sense on its own.

44. REPLY GUIDE
Here are common comment types we get [paste examples]. Write
a short reply guide: tone, what to say, what never to say,
and when to escalate to a person.

45. EXECUTIVE GHOSTWRITING
Here are notes from [executive] [paste] and three of their
past posts [paste]. Draft a post in their voice. Keep their
opinions; don't add new ones.

#11Analysis and reporting prompts

46. GA4 EXPORT READ
Here's a GA4 export [upload]. Which pages and channels
changed most against the previous period? Show numbers,
separate real changes from noise, and list three questions
I should investigate.

47. MONTHLY REPORT DRAFT
From these numbers [paste or upload], draft a one-page
marketing report: three headline results, what drove them,
what didn't work and next month's priorities. Don't explain
causes the data doesn't show; mark guesses as guesses.

48. SURVEY ANALYSIS
Here are open-text survey answers [upload, anonymised].
Group them into themes with counts and quotes. Note any
theme mentioned by fewer than five people as tentative.

49. FORECAST SANITY CHECK
Here's our pipeline forecast [paste]. What assumptions does
it rely on, which look optimistic, and what would change the
number most?

50. RETRO
Here's what we planned for [campaign] and what happened
[paste]. Run a short retrospective: what worked, what didn't,
what we'd change and what to keep doing.

#12How do you turn these into a team library?

Copying prompts into a shared document is a good start, but documents get stale and people stop opening them. Move the prompts you use weekly into Claude itself.

Projects suit prompts tied to a body of knowledge. Put the brand voice guide, fact sheet and your three most-used prompts in the project instructions, and everyone working in that project gets them.

Skills suit repeatable tasks. A Skill is a folder with a SKILL.md file that starts with a name (up to 64 characters) and a description (up to 200 characters), followed by instructions. Claude reads the descriptions and loads a Skill when your request matches, so the description should say exactly when to use it (Anthropic). Skills are available on Free, Pro, Max, Ultra, Team and Enterprise plans and need code execution turned on (Anthropic).

A minimal Skill for prompt 16 looks like this:

---
name: Fact check list
description: Use when asked to fact-check, verify or review
  claims in a marketing draft before approval.
---
List every factual claim in the draft. For each, say whether
it is supported by the fact sheet in this project, by a source
cited in the draft, or by nothing. Present a table with
columns: claim, support, source. Do not rewrite the draft.

To install it, put the file in a folder, zip the folder so the folder is at the root of the ZIP, and upload it under Customize > Skills. On Team and Enterprise plans you can share a Skill with named colleagues or groups; it appears in their list switched off until they turn it on (Anthropic).

#13How do you keep the library good?

Give each prompt an owner and a last-reviewed date. Once a month, spend 30 minutes on it:

• Delete prompts nobody used.
• Rewrite prompts whose outputs keep needing the same fix. That fix belongs in the prompt.
• Add prompts people wrote for themselves and shared in chat.
• Check prompts that rely on facts, such as character limits or prices, are still right.

Ask Claude to help with the review:

Here are our library prompts [paste] and notes on what went
wrong with recent outputs [paste]. For each prompt that had
problems, suggest a specific change to the prompt. Don't
change prompts that worked.

#14Worked example: Ledgerline's first library

Ledgerline is a fictional accounts-payable tool. Details are illustrative.

Ledgerline's team started by listing the work they did every week: blog drafts, LinkedIn posts, a monthly report and search ads. They picked eight prompts from the list above (2, 11, 12, 16, 21, 32, 41 and 47), replaced the brackets with standing details, and saved them with one good output each.

Three became Skills, because people ran them several times a week: the fact check list, the repurposing prompt and the search ad assets prompt. The others went into the content project's instructions. After the first monthly review, they deleted one prompt nobody used and added a tighter rule to prompt 12 after drafts kept running long.

#15How does this connect to the certificate?

The track's optional project, "MentionX Certified: Claude for Marketers", asks for two things: a prompt library of at least ten prompts you actually use, each with purpose, context and an example output, and one automated workflow. This lesson gives you the raw material for the first. For the second, lessons 12 to 14 cover scheduled tasks, sheets and connectors. The track overview explains what to submit.

Next: Measuring your brand's visibility in Claude answers. Or return to the Claude for Marketers track.

#16Common questions

Where should a marketing team keep its Claude prompts? Start with a shared document, then move the prompts you use weekly into a Claude project or a Skill so they load automatically. Skills can be shared with colleagues on Team and Enterprise plans.

Will these prompts give the same result every time? No. Claude's answers vary between runs, and results depend on the context you add. Treat each prompt as a starting structure, and save a good example output next to it so people know what to aim for.

How many prompts does a team actually need? Fewer than you think. Most teams rely on ten to twenty prompts for their recurring work. The rest of this list is there to pick from, not to keep in full.