Templates

AI Prompt Library for B2B Marketing Teams (Free + Copy-Paste Ready)

9 min read · Jun 14, 2026· AO Network Editorial Team

Most AI prompt libraries online are 200-prompt Notion dumps. Most of them produce generic output because the prompts are written by someone who has never run a B2B marketing program.

The version below is shorter, sharper, and tuned for B2B marketing teams. Eight prompts across the work most marketing operators do every week. Copy any of them directly from this page. The button is in the top-right of each prompt card.

If you have not picked an AI tool yet, the best AI marketing tools post covers the platforms these prompts work in.

How to actually use these

Replace anything in [square brackets] with your specific context. Paste the filled-in prompt into Claude, ChatGPT, or Gemini. Edit the output. Ship.

The prompts assume you are running an always-on marketing program. They reference an ICP, a brand voice, and a content calendar. If you do not have those, the ICP and persona worksheet and the content calendar template are the prerequisites.

Content drafting

Prompt

B2B blog post first draft

Recommended model: Claude Sonnet 4.7 or GPT-5

You are an experienced B2B marketing writer with strong opinions and a skeptical voice. You write for [VP of Marketing at mid-market SaaS companies, 100 to 1,000 employees].

Write a 1,200-word blog post on [TOPIC].

Voice rules:
- Short paragraphs (2 to 3 sentences max)
- Lead with a sharp observation, not background
- Use specific examples, not abstract claims
- One opinion per section
- End with a question that invites response
- No corporate jargon (no synergy, leverage, unlock, holistic)
- No em dashes; use commas or periods

Structure:
- Hook (1 paragraph) that names the problem in plain language
- 5 to 6 H2 sections, each starting with a sentence-form claim
- A short FAQ section with 3 questions
- A closing question to the reader

Do not use the word 'just' as filler. Do not use rule-of-three lists by default. Vary sentence length.

Return the full draft. Avoid filler sentences.

Prompt

LinkedIn post in 4 angles

Recommended model: Claude Sonnet 4.7 or GPT-5

I will paste an article below. Generate four LinkedIn post drafts that each take a different angle on the same article.

The four angles:
1. The counter-take. Argue against the common wisdom in the article.
2. The personal anecdote. Open with a specific moment from operator experience that the article maps onto.
3. The list breakdown. Three to five tight bullets from the article, no filler.
4. The question post. Lead with the question the article tries to answer, then give a one-paragraph answer that ends with 'curious how others see it'.

For each:
- Open with a hook in the first line that works without context
- Keep to 180 to 250 words
- No emojis, no hashtag spam (max 2 hashtags)
- End with a question to invite replies

Article:
[PASTE FULL ARTICLE TEXT HERE]

Customer research

Prompt

Extract ICP patterns from interview transcripts

Recommended model: Claude Sonnet 4.7

I will paste 5 customer interview transcripts below. Each is a 30-minute conversation with a recent customer.

Your job is to extract patterns that inform our [ICP and persona worksheet](/templates/icp-persona-worksheet-template).

Return:
1. Trigger patterns: what was happening at each company in the 90 days before the buying decision? List the specific triggers, with frequency counts.
2. Job-to-be-done patterns: what were these buyers measured on? List the specific metrics, with frequency counts.
3. Information sources: where did they look for information during their search? List specific publications, podcasts, people, conferences, with frequency counts.
4. Disqualifier signals: from the transcripts, what would have disqualified each company even if firmographics had matched? Be specific.

For each finding, cite which transcript(s) supported it (e.g., 'Transcripts 1, 3, 4').

Do not invent patterns that appear in only one transcript unless they are unusually specific. Do not generalize beyond the data.

Transcripts:
[PASTE TRANSCRIPT 1]
---
[PASTE TRANSCRIPT 2]
---
[PASTE TRANSCRIPT 3]
---
[PASTE TRANSCRIPT 4]
---
[PASTE TRANSCRIPT 5]

Prompt

Voice-of-customer mining from sales calls

Recommended model: Claude Sonnet 4.7 or GPT-5

I will paste 10 sales call transcripts below. These are first conversations with prospects.

Return:
1. The 5 most common pain points mentioned, in the prospect's exact language
2. The 3 most common objections, with the specific phrasing used
3. The 5 most common questions prospects ask about us, in their words
4. Three specific phrases that came up multiple times that should appear in our marketing copy verbatim

For each item, give exact quotes with the transcript number.

Flag anything that contradicts our current positioning.

Transcripts:
[PASTE 10 TRANSCRIPTS, SEPARATED BY ---]

Briefs and planning

Prompt

Marketing brief reviewer

Recommended model: Claude Sonnet 4.7

Below is a marketing brief I wrote. Review it against the [seven-field marketing brief template](/templates/marketing-brief-template) and flag any of the following:

1. Vague outcomes (anything that is not a measurable result)
2. Vague audiences (anything broader than a specific role at a specific company type)
3. Missing or weak success metrics
4. Constraints that are actually preferences
5. Deliverables that are not enumerated
6. Missing decision-maker or reviewer

For each issue, name the field and explain what is wrong in one sentence. Then suggest a specific rewrite.

Do not soften your critique. The brief gets stronger when problems are surfaced.

Brief:
[PASTE BRIEF HERE]

Prompt

Quarterly plan stress test

Recommended model: Claude Sonnet 4.7

I will paste my draft [quarterly marketing plan](/templates/quarterly-marketing-plan-template) below.

Act as a skeptical CFO reviewing it for the first time. Push back on:

1. The quarter goal: is it specific enough to know whether we hit it?
2. The always-on layer: which channel listed here could we cut in half without hurting the goal?
3. The campaign layer: which campaign is competing with the always-on layer for attention?
4. The bets: do any of them lack a clear kill condition?
5. The 13-week cadence: where is the team most likely to slip?
6. The risks and reserves: what risk is missing that should be on the list?

Be blunt. Give a one-line verdict at the end: ship as is, ship with edits, or rebuild.

Plan:
[PASTE PLAN HERE]

Analysis

Prompt

Competitor positioning extractor

Recommended model: Claude Sonnet 4.7 or GPT-5

I will paste the homepages and main solution pages of three competitors below.

For each competitor, extract:

1. Positioning statement in one sentence (their primary claim, in their own language)
2. ICP they appear to target (specific company size, industry, role)
3. The 3 most prominent benefit claims, in their exact words
4. Their pricing model and visible price points
5. The objections they explicitly handle on the page
6. Notable absences: what would you expect a competitor in this space to say that they do not say?

Then, do a brief comparison: what is the single positioning angle where each competitor is strongest? Where do all three agree (which is where we should not try to compete)? Where is there whitespace?

Competitors:
Competitor A:
[PASTE HOMEPAGE + SOLUTION PAGE TEXT]
---
Competitor B:
[PASTE]
---
Competitor C:
[PASTE]

Prompt

Cold outbound email writer

Recommended model: Claude Sonnet 4.7 or GPT-5

Write a 4-email outbound sequence to [TARGET PERSONA: VP of Marketing at mid-market B2B SaaS].

Product context: [DESCRIBE WHAT YOU SELL IN ONE PARAGRAPH].

Most recent trigger event for these accounts: [HIRING SIGNAL / FUNDING ROUND / LEADERSHIP CHANGE - BE SPECIFIC].

Voice: peer-to-peer, no marketing language, no spam tells, never promise revolutionary results.

Sequence rules:
- Email 1 (day 1): 90 words max. Reference the trigger. State a specific reason it makes our solution relevant now. Soft CTA (one question, no calendar link).
- Email 2 (day 4): 60 words. Add one specific data point or customer outcome that is relevant to their situation.
- Email 3 (day 9): 50 words. Forward-style with FW: subject. Quote one line from a customer story that mirrors their situation.
- Email 4 (day 14): 40 words. The breakup. State that you will stop reaching out. Leave the door open.

No subject lines longer than 5 words. No 'just checking in' subject lines. No 'I noticed you'.

Return all four emails formatted with subject lines.

What to skip

AI prompts that promise to generate full marketing strategies. The strategy is the judgment call. AI can structure a plan after you have decided on it. It cannot do the deciding.

Prompts that ask AI to 'be a marketing expert'. The role-play framing does not improve output quality and often makes it worse. Be specific about what you want, not who the AI is.

Prompts that ask for output in a specific length without specifying voice constraints. You will get generic content stretched to the requested word count.

How this fits with the broader always-on motion

These prompts are accelerators on top of a working program. Without an ICP, a calendar, a brief template, and a brand voice, AI accelerates the wrong work faster. The marketing audit framework is the right starting point if your foundation is shaky.

Frequently asked questions

Which AI model should I use for these?

Claude Sonnet 4.7 produces the strongest B2B marketing output today. GPT-5 is close. Gemini handles bulk processing well and is the right pick when token cost matters. The differences between the three are small enough that picking the model your team already uses beats switching tools.

Should I use these prompts inside a marketing automation tool?

Some of them, yes. Customer.io, Klaviyo, and HubSpot all support AI-generated content inside flows. The customer research and analysis prompts work better in a standalone chat interface where you can iterate.

How often should I update prompts?

Every quarter. AI models improve and prompts that worked six months ago often work better with small tweaks. Treat the prompt library like the content calendar: living, not static.

Which prompt above is closest to a workflow your team runs manually right now? That is the one to copy first.

The Always-On Brief

Weekly strategy, tool picks, and playbooks. 6,000+ marketers subscribed.

Templates

AI Customer Research Prompt Pack (Free B2B Framework for Extracting Real Insights)

Jun 16, 2026

Templates

AI Content Brief Template (Free Framework for Briefing AI Tools to Match Your Brand Voice)

Jun 15, 2026

Templates

Always-On Marketing KPI Dashboard Template (Free + How to Use It)

Jun 14, 2026