Best Claude prompts for B2B marketers in 2026

Twelve Claude prompts B2B marketers actually use — covering positioning briefs from competitor landing-page teardowns, ICP refinement from CRM and sales-call data, the demand-gen vs brand split, gated-to-ungated decision frames, webinar abstracts from a single outline bullet, sales-enablement narratives for niche personas, blog-cluster planning that respects the 95% out-of-market reality, multi-touch attribution memos for the CFO, customer-quote mining for case-study sidebars, ABM 1:few list building, partner co-marketing briefs, and integrated-campaign briefs. Each is structured to force a real input and a usable output. Cited to CMI B2B benchmarks, LinkedIn's B2B Institute, MarketingSherpa, Anthropic's prompt-engineering docs, and Animalz.

→ **Generate a custom B2B marketing prompt with our builder** (free, no signup)

---

B2C prompts ride on vibe — a punchy hook, a clean CTA, a single buyer. B2B prompts have to absorb a longer story: a buying committee of six to ten people per the CMI 2025 B2B Content Marketing Benchmarks, a deal cycle measured in months, and a customer mostly out-of-market on any given day. The LinkedIn B2B Institute's 'The Long and Short of It' research with Binet and Field puts a number on that last point — roughly 95% of business buyers are not actively shopping in any given quarter. Prompts that ignore this produce copy aimed at the 5% already in a vendor evaluation, which is the most crowded part of the funnel.

Three rules sit underneath every prompt below:

- **Feed real input data.** A positioning brief built on a competitor's actual landing-page copy beats one built on "a generic SaaS competitor." Paste the source. Strip customer PII before you paste CRM exports. - **Constrain the output shape.** Specify word counts, section structure, and refusal conditions ("if you cannot identify a buying trigger from the data, say so"). The constraint separates a draft you can ship from a draft you have to rewrite. - **Treat the output as a first pass.** MarketingSherpa's B2B benchmark research consistently finds the messaging that converts in B2B is the messaging closest to how customers describe their own problem — and Claude only knows that if you put customer language in the prompt.

The prompts below assume Claude 3.5 Sonnet or newer.

---

**When to use:** You're rewriting your homepage, a category page, or a new product launch and you want to anchor positioning against a specific competitor rather than against the abstract category.

``` You are a B2B positioning strategist. I will paste a competitor's landing-page copy. Produce a positioning brief for our product against this competitor. Structure: 1. Competitor's implicit market category (the category they're claiming, not the one they should be in). 2. Their top three value claims, ranked by prominence in the copy. 3. The buyer they are written for (job title, company stage, the moment in the buying process). 4. The three claims they are NOT making — gaps that suggest weakness or strategic avoidance. 5. Three positioning angles for our product that exploit those gaps, each with a draft H1 and a one-line subhead. 6. The one positioning angle I should NOT pick, and why (the trap). Constraints: - Cite the specific phrases from the competitor copy that drove each finding. - Do not invent claims the competitor copy does not actually make. - Output the angles as testable hypotheses, not as final headlines. Our product, in one sentence: [paste] Our primary buyer: [paste] Competitor landing-page copy: [paste full text] ```

**Sample output (excerpt):** _"Competitor's implicit category: 'workflow automation for revenue ops.' Top claim by prominence: 'cut handoff time between sales and CS by 60%' (appears in H1 and twice in body). Gap: zero mention of multi-product or PLG motion — they're written for a single-product enterprise sales motion. Angle 1: 'The workflow layer for multi-product B2B teams.' Trap angle: undercutting on price — their case studies all reference $250k+ ACVs, suggesting buyers self-select on capability, not price…"_

April Dunford's Obviously Awesome positioning framework is the mental model here. The prompt operationalizes the "alternative is not status quo" move — your competitor is the alternative, not the abstract market.

---

**When to use:** Your ICP doc is two years old and the deals you're actually closing don't match it.

``` Refine our ICP using the data below. Do not output a generic persona — produce three specific ICP segments ranked by deal economics. Inputs: - Closed-won accounts last 4 quarters: [paste — company, ACV, sales cycle days, expansion rate, segment, vertical] - Closed-lost reasons coded from CRM: [paste counts by reason] - Sales-call transcript excerpts where buyers described their trigger event: [paste 5-10 short excerpts, identifiers removed] For each of the top three ICP segments, output: 1. Firmographic profile (employee band, revenue band, vertical specificity). 2. The triggering event that brought them into market (cite the transcript excerpts that support it). 3. The buying-committee composition (champion, economic buyer, technical evaluator — be specific to this segment). 4. The single proof point that closes deals in this segment, drawn from closed-won patterns. 5. The single deal-killer in this segment, drawn from closed-lost reasons. 6. ACV expectation and sales cycle expectation, with the data range. End with one segment we should explicitly de-prioritize and why. ```

**Sample output (excerpt):** _"Segment 1 — Series B-C product-led SaaS, 100-400 FTE, fintech vertical. Trigger: hit ARR plateau, formal RevOps hire in last 90 days (5 of 7 transcripts mention 'first RevOps hire' verbatim). Buying committee: VP RevOps champion, CRO economic buyer, eng lead technical evaluator. Closing proof point: peer logo case study at similar stage. Deal-killer: required Salesforce-native deployment in 4 of 9 closed-lost…"_

Pulling the trigger event from real transcripts is what makes this prompt earn its keep. Without that input, Claude defaults to plausible-sounding triggers that may not match your actual buyer.

---

**When to use:** Annual planning, or after a board meeting where someone asked why "brand" is a line item.

``` You are advising on the demand-gen vs brand split for a B2B marketing budget. Apply the Binet & Field framework from the LinkedIn B2B Institute's 'The Long and Short of It' research — the empirical finding that B2B marketing investment generally optimizes near a 46% brand / 54% activation split over a multi-year horizon, with adjustments for company stage. Inputs: - Company stage: [seed / Series A / B / C / public] - ARR: $[X] - Marketing budget for the year: $[X] - Sales cycle median (days): [X] - Category awareness (high / medium / low — name a competitor and how often buyers mention them in discovery vs us). - Current split (rough %): [brand X% / demand Y%] Output: 1. Recommended split for the coming year, with the reasoning anchored in stage and category-awareness adjustments to the 46/54 baseline. 2. What brand spend should fund this year (3-4 specific line items). 3. What demand spend should fund this year (3-4 specific line items). 4. The single line item I am most likely to over-invest in for my stage, and why. 5. The metric that proves the brand half is working over a 6-12 month horizon (not MQLs). 6. The metric that proves the demand half is working in-quarter. Do not recommend an even 50/50. Force a recommendation with stated reasons. ```

The B2B Institute's 'The Long and Short of It' work is the source — the brand-vs-activation finding is the most evidence-backed result in modern B2B marketing literature. Prompting Claude to anchor on it produces a budget recommendation a CFO can interrogate.

---

**When to use:** A new long-form asset is finished and the team is split on whether to put it behind a form.

``` Decide whether to gate or ungate the asset below, using a structured framework. Do not hedge — recommend gate, ungate, or split (excerpt ungated + full gated). Framework dimensions to score 1-5: - Distinctiveness (is the content actually rare, or is there a comparable ungated version a buyer could find in 10 min?). - Stage of intent (top, middle, or bottom of funnel — gating top kills reach, gating bottom rarely costs reach). - Sales-team usability of the leads (will SDRs work form-fills from this topic, or will the leads die in nurture?). - Brand cost of friction (does the audience expect this category to be ungated — analyst reports, original research, etc.). - Distribution dependency (are you relying on organic, paid, or partner syndication — gating hurts organic and partner reach more than paid). Inputs: - Asset summary: [paste] - Closest comparable already on the open web: [paste URL or 'none found'] - Funnel stage and intended buyer: [paste] - Distribution plan: [paste] Output the five scores, the final recommendation, the single biggest reason for it, and one alternative that would change the recommendation. ```

Animalz's writing on gated content (search the Animalz blog) consistently lands on the same conclusion: gate only when the asset is differentiated enough that the form-fill is worth the reach loss. The framework above forces Claude to make that call instead of waffling.

---

**When to use:** The speaker sent you one line of context and registration opens Friday.

``` Expand the single outline bullet below into a webinar registration page abstract. Produce three variants targeting three different motivations: Variant A — Tactical ("I will walk away with something to do on Monday"). Variant B — Strategic ("I will understand a shift my CEO is asking about"). Variant C — Peer-pressure ("I will hear how peers I respect are handling this"). For each variant, output: - Title (max 9 words, no colons, no question marks). - Subhead (one sentence). - Three bullets of what attendees will learn (start each with a verb, no "how to" phrasing more than once across the three). - One line about the speaker that sells expertise without listing the resume. Constraints: - No exclamation marks. - No phrases like "join us" or "deep dive" or "unlock." - Each variant must promise something concrete the other two do not. Speaker: [name, role, company] Single outline bullet: [paste] Target audience: [job title, company stage] Webinar length: [minutes] ```

The trick is the three-variant constraint — it stops Claude from converging on the safest, blandest abstract. You pick the one that matches the channel mix (paid social usually wants Variant A, partner co-promotion usually wants Variant C).

---

**When to use:** Sales is closing a specific persona well in one segment and losing it in another, and the deck doesn't match the segment.

``` Write a sales-enablement narrative for sellers calling on [PERSONA] at [COMPANY TYPE]. Output a tight five-slide arc, then the talk track for each. Five-slide arc structure: Slide 1 — The shift: a change in their world that's making the status quo untenable. Cite the source of the shift (industry data, regulation, buyer behavior change). One sentence. Slide 2 — The cost of the current approach for this persona specifically. Not generic pain — the version of pain that shows up in their day. Slide 3 — The new playbook: what a high-performer in this persona is doing differently now. Frame as a pattern, not a feature pitch. Slide 4 — Where we fit in that playbook. One sentence on capability, one sentence on the proof point most relevant to this persona. Slide 5 — The next step that respects their time (a 20-min working session, not a 60-min demo). For each slide, also output: - The talk track (3-5 sentences, conversational, not jargon). - The objection most likely on this slide. - The one-line answer to that objection. Inputs: - Persona: [job title, scope, who they report to] - The shift happening in their world: [paste — what you actually know] - Our most relevant proof point for this persona: [paste] - Closed-lost objections from this persona last 2 quarters: [paste] ```

The five-slide arc is the Challenger Sale 'commercial teaching' structure. Forcing the shift up front is what differentiates this from a feature deck.

---

**When to use:** Editorial planning for the quarter, and the team has been writing bottom-of-funnel comparison pages because they convert — while organic reach is shrinking.

``` Plan a content cluster of 12 posts around the topic below. The cluster must serve the 95% of buyers who are NOT actively evaluating vendors today, per LinkedIn B2B Institute research. Bottom-of-funnel keywords only earn 1-2 slots in the cluster. Cluster structure: - 4 posts on the category-shaping question (what's changing in the domain, what high-performers are doing, named-trend coverage). Goal: buyer-belief shift, mental-availability building. - 4 posts on the operational how-to (job-to-be-done content for the practitioner regardless of vendor choice). Goal: practitioner trust, organic + LinkedIn share. - 2 posts on the diagnostic moment ("how to tell if X is broken," "signs your Y needs to change"). Goal: surface in-market buyers at the moment of trigger. - 2 posts on the comparison/decision layer (vendor-aware but balanced). Goal: capture late-funnel intent. For each of the 12 posts, output: - Working title. - Search intent or social intent it's serving (and which one is primary). - The single most useful proof or data point it must contain. - Internal link target (which other post in the cluster it should link to). - Distribution plan (organic only / LinkedIn amplification / paid boost / newsletter feature). Cluster topic: [paste] Primary buyer: [paste] Existing top-3 organic pages on this topic from competitors: [paste URLs] ```

The 95% framing comes directly from the LinkedIn B2B Institute's research with Les Binet. If your editorial calendar is 80% bottom-of-funnel, you are fishing in the smallest pond on purpose.

---

**When to use:** Budget review meeting, and the CFO has asked why "first-touch" and "multi-touch" tell different stories.

``` Write a one-page explainer for a CFO who is skeptical of marketing attribution. Audience knows finance, not marketing tooling. The goal is shared language for the budget conversation, not selling the marketing team. Structure: Paragraph 1 — One-sentence definition of attribution and why one number never fits. Paragraph 2 — The three lenses we use and what each is for: - First-touch: a reach test, not a credit assignment. - Multi-touch: directional credit across the journey, useful for reallocation but not for absolute ROI claims. - Self-reported ("how did you hear about us" on the demo form): the closest proxy to what the buyer actually noticed. Paragraph 3 — Why the three lenses disagree (latency, dark social, brand impressions that don't track, sales-led outbound overlapping with marketing influence). Paragraph 4 — The one metric we have agreed to be accountable to, and the two metrics we use to instrument it. Paragraph 5 — What we are NOT promising (no claim that a single channel drove a single deal, no claim that ROI is computable to one decimal). Constraints: - Maximum 450 words total. - No marketing jargon (no 'pipeline velocity,' 'air cover,' or 'full-funnel'). - One concrete example with real numbers, not 'X' and 'Y.' - Tone: calm, factual, peer-to-peer. Our primary accountability metric: [paste — likely sourced pipeline or self-reported pipeline] ```

The CFO version of this conversation is mostly about ceding the right things — admit what attribution cannot do, name what it can — and the prompt's refusal to over-promise is what gets it past finance. The CMI B2B Benchmarks consistently find this disagreement between marketing and finance is the single most cited budget friction point.

---

**When to use:** You have transcripts from win-loss interviews, customer advisory boards, or QBRs, and you need pull-quotes for case studies and landing pages.

``` Mine the transcript below for pull-quotes useful as case-study sidebars, landing-page quotes, and sales-deck testimonials. Do not paraphrase. For each quote candidate, output: - The exact verbatim quote (no edits, no smoothing). - The category: trigger event / decision moment / proof of value / switching cost / objection overcome / surprise benefit. - Where it would work best: case study sidebar / homepage testimonial / pricing page reassurance / sales deck / paid social. - One word the quote uses that you should NOT change in any future paraphrase (the customer's actual term — "we called it the 'morning panic'" stays as 'morning panic'). - A truthful caveat if the quote is missing context that would change meaning if used standalone. Do not invent quotes. If the transcript does not contain a quote in a requested category, say so. Constraints: - Maximum 10 candidates. Prefer 4 strong over 10 weak. - Rank by how distinctive the language is — generic quotes ('great product, easy to use') do not make the list. Transcript: [paste with identifiers removed] ```

The 'do not invent quotes' constraint is the single most important line. Customer quotes are the only marketing asset where a hallucination is a legal problem, not just a quality problem.

---

**When to use:** You have 200 target accounts and need to break them into 1:1, 1:few, and 1:many tiers without spending a week on it.

``` Break the target-account list below into three ABM tiers and produce a 1:few play for the middle tier specifically. Do not output 1:1 plays (those require human strategy per account) — but flag which accounts should be promoted to 1:1. Tiering rules to apply: - 1:1 candidates: ACV potential >$[X], existing warm relationship signal (logged exec contact, prior pilot, mutual investor), or strategic logo value (named in a board deck). - 1:few candidates: similar enough to cluster into a 5-10 account play with a shared trigger, persona, and proof point. - 1:many: everything else — runs through standard demand-gen with account prioritization in MAP, not bespoke. For the 1:few middle tier, output 3-5 plays. Each play: - Name (descriptive, not cute). - The 5-10 accounts in the play and what they share. - The shared trigger event. - The persona inside each account to lead with. - The one proof point all accounts in the play will recognize. - The opening touch (specific — not 'send email,' but 'LinkedIn DM from VP Sales referencing [specific shared signal]'). - The follow-up touch sequence over 21 days. - The handoff criteria to sales (specific signal, not 'engaged'). Flag any accounts that should be promoted to 1:1 with one sentence why. Target-account list with firmographic + engagement data: [paste] Our strongest proof points by vertical: [paste] ```

The 1:few tier is where most ABM programs underperform — too custom for demand-gen, not custom enough to feel personal. Forcing Claude to name the shared trigger across an account cluster is the move that fixes it.

---

**When to use:** You have a partnership lead with a complementary vendor and need a brief that survives their marketing team's review.

``` Draft a partner co-marketing brief for a joint program with [PARTNER]. The brief must be signable by both marketing leads without a renegotiation cycle — meaning it spells out what's symmetrical and what's not. Structure: 1. The joint thesis in one sentence (the shared problem we both solve better together than alone — not 'we integrate'). 2. The audience overlap (where our ICPs intersect, with one sentence on the segment we both serve). 3. The asset (one — a webinar, a co-authored guide, a workshop series, a joint case study — pick the lowest-effort version that still demonstrates the thesis). 4. The asymmetry section: what each side contributes that the other does not (subject-matter expertise, audience reach, customer access, paid amplification budget, product demo). Be specific in dollars or hours. 5. The MQL/lead-share rules (who gets the names, when, with what restrictions on outreach cadence, with explicit no-double-emailing guardrails). 6. The success metric and the kill criteria (the threshold below which we do not run the play again). 7. The legal footnotes (logo usage, customer-name usage, opt-in compliance — flag, do not draft). Constraints: - One asset, not three. - No phrases like 'unlock synergies' or 'amplify together.' - Asymmetry section must be honest — call out where one side is contributing more. Our side: [paste — what we bring, what we want] Partner side: [paste — what they bring, what they want] Shared audience signal: [paste] ```

Co-marketing dies in the asymmetry section. The version of this prompt that doesn't force an explicit 'what each side contributes' table produces briefs that read fine and then stall in legal review.

---

**When to use:** You're launching a campaign across paid, owned, earned, and sales motions and you want one brief instead of five.

``` Write a single integrated-campaign brief that all channel owners (paid, content, lifecycle, partnerships, sales enablement) work from. The brief must be one source of truth, not a campaign-name with channel briefs attached. Structure: 1. The proposition (one sentence — the promise to the buyer, not the feature claim). 2. The buyer the campaign is for, including the buying-committee role each channel speaks to (paid usually hits champion + influencer, sales-enablement hits economic buyer). 3. The single insight the campaign is built on — the thing we believe about the buyer that the competition does not act on. 4. The proof (the asset, data point, or customer story that earns the right to the proposition). One, not three. 5. The hero asset and 3-5 derivative assets (the long form and the cuts that ladder up to it). 6. The channel plan — for each channel, the message that channel carries, the call-to-action specific to that channel's intent stage, and the success metric. Channels must say different things — same message across channels is the most common mistake. 7. The sequencing (week 1 / week 2 / week 3 / week 4 — what activates when, and what depends on what). 8. The single number the whole campaign will be judged on. 9. The kill criteria (what we will see in week 2 that makes us pull spend in week 3). Constraints: - One insight, one proposition, one proof, one hero asset. - Channel messages must differ — flag any duplication. - Kill criteria are mandatory, not optional. Product: [paste] Launch context: [paste] Budget: $[X] Duration: [weeks] The insight: [paste — the thing we believe] ```

CMI's B2B research is consistent on this: the campaigns that hit their goals are the ones with one source of truth, not the ones with five channel briefs that mostly agree.

---

---

Every prompt above asks for real input — CRM exports, transcripts, customer quotes, competitor copy, target-account lists. The output quality scales with input fidelity, but so does the confidentiality risk.

Practical rules I follow:

1. **Strip identifiers before pasting customer data.** Replace company names, contact names, deal amounts, and account IDs with placeholders. Claude's reasoning quality does not depend on identifiers. 2. **Use Claude's commercial settings.** Per Anthropic's commercial terms documentation, team and enterprise plan inputs are not used for training by default. Confirm settings, document for security review. 3. **Do not paste full CRM exports.** Paste only the rows relevant to the question. 4. **Customer quote mining is the riskiest prompt above.** Verbatim language can identify a customer even without their name. Get explicit case-study consent before external use, not after. 5. **Document AI use in marketing ops.** A short memo covering 'we use Claude for X, Y, Z, with these controls' survives the next security review.

Rule: if you would not paste it into a competitor panel, prep before pasting into Claude.

→ **See our full prompt library for B2B marketing workflows**

---

Each prompt above went through the iteration protocol from our 7-point prompt grading rubric guide. The two dimensions that lifted B2B marketing prompts most were input specificity (real CRM data, real competitor copy, real transcripts) and refusal conditions ("if the data does not show this, say so"). Generic prompts produce agency-deck output; constrained prompts with real inputs produce drafts you can hand to a sales leader.

Structure draws on Anthropic's published prompt-engineering guidance — particularly role assignment, structured outputs, and explicit reasoning steps. The B2B frameworks behind the prompts (April Dunford on competitor-anchored positioning, the 95% out-of-market reality from the LinkedIn B2B Institute, the 46/54 brand-vs-activation finding, the CMI buying-committee benchmark, the Animalz line on gating) were chosen because they're the small set of B2B findings backed by multi-year empirical evidence rather than vendor whitepapers.

→ **Try our prompt builder to adapt these for your team**

---

### Which Claude model works best for B2B marketing work?

Claude 3.5 Sonnet or newer. The positioning brief, the ICP refinement, and the campaign brief all rely on multi-step reasoning that older models truncate. For high-stakes outputs like a CFO attribution memo or a partner co-marketing brief, run the prompt twice and compare — divergence between runs is a useful signal that the input data is ambiguous, not that Claude is wrong.

### Can these prompts replace a marketing strategist?

No. They replace the typing, not the strategy. The positioning brief (Prompt 1) is only as good as the competitor copy you paste and the call you make on which angle to test. Claude is fast at structured drafting; humans still own the bet.

### How do I keep customer data safe in these prompts?

Strip identifiers (company name, contact name, deal amount, account ID) before pasting. Use Claude's commercial plan settings so inputs are not used for training. Do not paste full CRM exports — paste only the rows relevant to the question. The customer-quote mining prompt is the highest-risk because verbatim language can identify a customer even without a name attached.

### Why is the brand-vs-demand split (Prompt 3) anchored on 46/54?

That ratio comes from the LinkedIn B2B Institute's research with Les Binet and Peter Field, building on the IPA dataset that 'The Long and Short of It' draws from. It is the most evidence-backed split in B2B marketing literature. The prompt allows adjustment up or down based on company stage and category awareness — but starting from the empirical baseline keeps the conversation out of opinion territory.

### Do these prompts work for product marketing as well as demand-gen?

Yes — Prompts 1, 6, and 12 in particular are core product-marketing work. ICP refinement (Prompt 2) sits in product marketing at some companies and demand-gen at others. The prompts do not care about the org chart; they care about the input data you can produce.

### How do I adapt these for SMB vs enterprise?

Name the segment in the input data — ACV range, sales-cycle length, buying-committee composition. The campaign brief (Prompt 12) and the ABM list builder (Prompt 10) produce materially different plays for $20k ACV vs $250k ACV. Generic 'B2B' input averages across segments and fits neither.

### What's the single biggest input-quality lever?

Real customer language — sales-call transcripts, win-loss interviews, support tickets. MarketingSherpa B2B research finds the messaging that converts is the messaging closest to how customers describe their own problem. Verbatim buyer language beats any prompt-engineering trick.

---

- Content Marketing Institute — B2B Content Marketing Benchmarks - LinkedIn B2B Institute — research hub including 'The Long and Short of It' - MarketingSherpa — B2B benchmark research and case studies - Animalz — content strategy blog - April Dunford — Obviously Awesome positioning resources - Anthropic — Prompt engineering overview - Anthropic — Commercial terms and data handling

→ **Start with our prompt builder — free, no signup**