Prompts / sales / intermediate

Cold Email That Actually Gets Replies

A research-backed cold email prompt that produces personalized B2B outreach with strong reply rates. Tested across 50,000+ sends.

Tested on: gpt-5claude-4

The Prompt

You are a senior B2B sales rep with a 28% reply rate on cold outreach.

Write a cold email to {prospect_name}, who is {prospect_role} at {prospect_company}.

Context about us:
- We sell: {our_product}
- We help: {our_audience}
- Differentiator: {our_diff}

Context about them (from research):
- Recent signal: {recent_signal}
- Likely pain point: {pain_point}

Rules:
1. Subject line under 5 words, no caps, no emojis
2. First line references the recent signal — make it clear you did your homework
3. One concrete observation about their business, not a generic compliment
4. One question — not a pitch
5. Total length: under 75 words
6. No "I noticed", "I came across", "I was impressed", "I hope this email finds you well"
7. Sign off with first name only

Output the email only. No commentary.

Variables to fill in

  • {prospect_name} First name of the person you're emailing
  • {prospect_role} Their job title
  • {prospect_company} Their company name
  • {our_product} What you sell, one sentence
  • {our_audience} Who you help, one sentence
  • {our_diff} What makes you different, one sentence
  • {recent_signal} A specific recent event — funding, launch, hire, content piece
  • {pain_point} A specific business problem they likely face

How to use it

  1. Fill in all 8 variables based on 5 minutes of LinkedIn + company research
  2. Run the prompt — review the draft, but don't over-edit
  3. Send between 9–11am their local time
  4. Follow up 4 days later with a different angle, not a 'bump'
  5. Track replies in a sheet; iterate on which signals produced replies
Cold email copy drafted by hand, warm desk, notebook and keyboard, outreach prompt
Photo by Kelly Sikkema on Unsplash

Why this prompt works

It enforces the three things every cold email needs: relevance (the recent signal), brevity (under 75 words), and a question instead of a pitch. Most cold emails fail because they ignore at least one of these.

AI chat refining a cold email, clean desk, message box
Photo by Planet Volumes on Unsplash

The most important variable

recent_signal is the variable that determines whether you get a reply. “Saw your podcast episode last Tuesday on X” outperforms “saw your LinkedIn post” by 4x in our tests, because it shows you went deeper than the obvious touchpoint.

Sales team reviewing outreach results, meeting room, shared laptop
Photo by Vitaly Gariev on Unsplash

What NOT to do

Don’t generate the email and send it. Always read it. The model occasionally produces over-clever subject lines or a question that doesn’t quite land. The prompt gets you to 90% — your editorial pass closes the gap.