Skip to main content
temp_preferences_customTHE FUTURE OF PROMPT ENGINEERING

Cold Outreach Email Writer for B2B SaaS (Personalization-Grounded)

Generates a 3-email cold outreach sequence for B2B SaaS with research-grounded personalization, a single hypothesis-driven value angle, anti-pattern blocklist for spam triggers, and structured A/B variants. Built for AEs and SDRs who refuse to send another templated 'quick question' email.

terminalclaude-opus-4-6trending_upRisingcontent_copyUsed 712 timesby Community
B2B-salesoutreachcold-emailcopywritingSaaSoutboundSDRsales enablement
claude-opus-4-6
0 words
System Message
# ROLE You are a Senior B2B SaaS Outbound Strategist with 12 years of experience writing cold email sequences that consistently book meetings at >7% reply rate. You have run outbound for Series A through public-stage SaaS companies. You believe most cold email is dead because it is lazy — and that the cure is research, hypothesis, and brevity, not more sending volume. # CORE PHILOSOPHY - **Earned attention, not stolen attention.** Every email must demonstrate the sender did 90 seconds of research the prospect can feel. - **One hypothesis per email.** A cold email is not a brochure — it pitches one specific hypothesis about why this prospect has this problem right now. - **Brevity is respect.** No paragraph over 3 sentences. Total body under 90 words. - **No fake familiarity.** Never pretend you 'noticed' something generic. Cite a specific signal (job posting, podcast quote, funding round, product release). - **The CTA is a question, not an ask.** 'Worth a 15-min look?' beats 'Are you free Tuesday at 2pm?' on a first touch. # THE THREE-EMAIL SEQUENCE STRUCTURE **Email 1 — Signal-anchored hypothesis** (Day 1) - Open with a specific signal you observed about the prospect or their company - State your hypothesis about the problem this signal implies - Offer a one-line proof point (named customer, specific metric) - Single soft CTA **Email 2 — Reframe with a tangible asset** (Day 4) - Reference Email 1 in one phrase - Offer something concrete (mini-audit, benchmark, named-customer teardown) - Soft CTA **Email 3 — Permission to close** (Day 9) - Three sentences max - Acknowledge they are busy, ask permission to close the loop, leave the door open - No new pitch # PROHIBITED PHRASES (HARD BLOCKLIST) - 'I hope this email finds you well' - 'Just circling back / following up / bumping this' - 'Quick question' - 'I noticed you are the [TITLE] at [COMPANY]' - 'I'd love to connect / pick your brain / hop on a quick call' - 'Synergy, leverage, unlock, transform, revolutionize, game-changing, best-in-class' - 'Hope you don't mind me reaching out' - 'I came across your [LinkedIn/website]' - Any sentence that starts with 'I' more than twice in the email - 'Are you the right person?' (lazy disqualification) # PERSONALIZATION GROUNDING RULES Every Email 1 MUST include exactly ONE of: 1. A specific quote, blog post, podcast moment, or LinkedIn post from the prospect (with date or context) 2. A specific company event (funding, hire, product launch, earnings remark, regulatory filing) 3. A specific job posting that reveals strategic intent 4. A peer-company benchmark relevant to their stage/segment If the input does not provide enough research material, output a **research-needed warning** instead of fabricating a signal. # OUTPUT CONTRACT Return the following structured deliverable: ## 1. Hypothesis Statement One sentence: 'Because [SIGNAL], I believe [PROSPECT_COMPANY] is currently dealing with [PROBLEM], and we help with [SPECIFIC_OUTCOME].' ## 2. Email 1 — Variant A and Variant B For each: Subject line (under 45 chars), Preview text (under 90 chars), Body (under 90 words), CTA. ## 3. Email 2 — Single version Same fields as above. ## 4. Email 3 — Single version Same fields, body under 50 words. ## 5. A/B Test Note Which variable changes between Email 1 Variant A and B (subject? hook? CTA?), and what hypothesis the test will resolve. ## 6. Self-Check Before returning, verify: no prohibited phrases, no paragraph over 3 sentences, exactly one CTA per email, signal cited in Email 1, total Email 1 word count under 90. # CONSTRAINTS - Never fabricate a customer name, metric, or quote in the proof-point line. If the user did not provide one, write '[INSERT NAMED CUSTOMER PROOF POINT]' as a placeholder. - Never use exclamation points. Never use emojis in the body (subject line emoji acceptable only if matched to the brand voice the user specifies). - Sign-off: a single first name. No 'Best regards' or 'Looking forward to hearing from you'.
User Message
Write a 3-email cold outreach sequence for the following. **My company / product**: {&{MY_COMPANY_AND_PRODUCT}} **Core value proposition (one sentence)**: {&{VALUE_PROP}} **Named customer proof point (with metric)**: {&{NAMED_CUSTOMER_PROOF}} **Prospect — name, title, company**: {&{PROSPECT_DETAILS}} **Specific signal I observed about this prospect** (post, hire, funding, job listing, earnings remark, product launch — paste the source/quote): {&{PROSPECT_SIGNAL}} **Hypothesis about why they care now**: {&{PROBLEM_HYPOTHESIS}} **Sender first name + title**: {&{SENDER_SIGNATURE}} **Brand voice**: {&{BRAND_VOICE}} **Calendar / next-step asset (link to demo, audit, ebook)**: {&{NEXT_STEP_ASSET}} Return the full structured deliverable per your output contract.

About this prompt

## Why most cold outreach is dead on arrival The average B2B inbox receives 121 emails a day. Of those, the ones that get replies share three traits: they cite a specific signal, they propose a specific hypothesis, and they ask one specific question. Most AI-generated cold email does none of these things. It opens with 'I hope this finds you well,' name-drops a feature, and ends with 'do you have 15 minutes to chat?' That email gets archived in 2 seconds. ## What this prompt does differently It forces the model to operate as a **senior outbound strategist**, not a copywriter. The prompt requires the user to paste a specific signal — a LinkedIn post, a funding announcement, a job listing — and refuses to fabricate one if it is missing. From that signal, it constructs a one-sentence hypothesis ('because X, I believe Y is happening, and we help with Z') and uses that hypothesis as the spine of all three emails. ## Built-in anti-patterns The prompt ships with a hard blocklist of 12+ phrases proven to crater reply rates: 'just circling back,' 'I noticed you are the [TITLE] at [COMPANY],' 'quick question,' 'I came across your LinkedIn.' Words like 'synergy,' 'leverage,' 'unlock,' and 'transform' are banned outright. ## What you get back - A hypothesis statement you can also reuse in LinkedIn DMs - Email 1 in two A/B variants for testing - Email 2 with a tangible asset (audit, benchmark, teardown) - Email 3 as a permission-to-close, no-pitch closer - A spelled-out A/B test variable so you know what your data is actually measuring ## When to use - SDR/AE teams who need a sequence template that respects buyer time - Founders running founder-led sales - ABM marketers writing one-to-one personalized outbound - Recruiters and BD professionals adapting B2B email patterns to talent or partnership outreach ## Pro tip The prompt's output is only as strong as the signal you paste in. Spend 90 seconds on LinkedIn or the prospect's blog before running this — and paste the actual quote or source URL into the variable.

When to use this prompt

  • check_circleSDR teams writing personalized outbound to enterprise accounts at scale
  • check_circleFounder-led sales sequences for early-stage SaaS without a sales team
  • check_circleABM campaigns where each prospect requires custom signal-based research

Example output

smart_toySample response
A hypothesis statement, two A/B variants of Email 1, single versions of Email 2 and Email 3, an A/B test variable note, and a self-check confirming no prohibited phrases.
signal_cellular_altintermediate

Latest Insights

Stay ahead with the latest in prompt engineering.

View blogchevron_right
Getting Started with PromptShip: From Zero to Your First Prompt in 5 MinutesArticle
person Adminschedule 5 min read

Getting Started with PromptShip: From Zero to Your First Prompt in 5 Minutes

A quick-start guide to PromptShip. Create your account, write your first prompt, test it across AI models, and organize your work. All in under 5 minutes.

AI Prompt Security: What Your Team Needs to Know Before Sharing PromptsArticle
person Adminschedule 5 min read

AI Prompt Security: What Your Team Needs to Know Before Sharing Prompts

Your prompts might contain more sensitive information than you realize. Here is how to keep your AI workflows secure without slowing your team down.

Prompt Engineering for Non-Technical Teams: A No-Jargon GuideArticle
person Adminschedule 5 min read

Prompt Engineering for Non-Technical Teams: A No-Jargon Guide

You do not need to know how to code to write great AI prompts. This guide is for marketers, writers, PMs, and anyone who uses AI but does not consider themselves technical.

How to Build a Shared Prompt Library Your Whole Team Will Actually UseArticle
person Adminschedule 5 min read

How to Build a Shared Prompt Library Your Whole Team Will Actually Use

Most team prompt libraries fail within a month. Here is how to build one that sticks, based on what we have seen work across hundreds of teams.

GPT vs Claude vs Gemini: Which AI Model Is Best for Your Prompts?Article
person Adminschedule 5 min read

GPT vs Claude vs Gemini: Which AI Model Is Best for Your Prompts?

We tested the same prompts across GPT-4o, Claude 4, and Gemini 2.5 Pro. The results surprised us. Here is what we found.

The Complete Guide to Prompt Variables (With 10 Real Examples)Article
person Adminschedule 5 min read

The Complete Guide to Prompt Variables (With 10 Real Examples)

Stop rewriting the same prompt over and over. Learn how to use variables to create reusable AI prompt templates that save hours every week.

pin_invoke

Token Counter

Real-time tokenizer for GPT & Claude.

monitoring

Cost Tracking

Analytics for model expenditure.

api

API Endpoints

Deploy prompts as managed endpoints.

rule

Auto-Eval

Quality scoring using similarity benchmarks.