Sales rep
ChatGPT Prompts for Sales Cold Emails
Also searched as: ChatGPT prompts for sales cold emails
- Who it is for
- SDRs, account executives, and founders who do their own prospecting and want personal-sounding emails without writing each one from scratch.
- Outcome
- A short, personalized first email plus a 3-touch follow-up sequence for one prospect, ready to edit and send from your own inbox.
Step-by-step prompts
Paste each step into the same chat so the model keeps the earlier context. Then edit before you send.
Step 1
Set up the offer
You're helping me write B2B prospecting emails. Here's what I sell: Product/service: [what it is] Who it's for: [ideal customer: industry, company size, job title] Problem it solves: [the problem in the buyer's words] Proof: [real customer results or facts I'm allowed to share] My ask: [e.g. a 15-minute call] Tone: plain, short, no buzzwords. Remember this for the rest of our chat.
Step 2
Personalize for one prospect
Prospect: [name], [title] at [company]. What I know about them: [recent news, LinkedIn post, job posting, tech they use; paste what you found] Write 3 opening lines that reference something specific from this info and connect it to the problem I solve. No flattery.
Step 3
Write the first email
Using the best opening line, write a cold email under [100] words. Structure: personal opener, the problem, one proof point, a low-pressure ask. Give me 3 subject lines under 6 words, lowercase is fine.
Step 4
Write follow-ups
Write 3 follow-ups for if they don't reply, sent [3, 7, and 14] days later. Each must add something new (a different angle, a short resource, a question) rather than "just checking in." Keep each under [60] words. The last one should be a polite close-the-loop email.
Step 5
Edit pass
Read the sequence as the prospect. What would make you delete it? Tighten anything that sounds salesy, generic, or too long.
Example of good output
EXAMPLE ONLY. Fictional prospect and company. Subject: your new ops hires Hi Dana, saw Northfield is hiring three ops coordinators. Usually that means the scheduling spreadsheet has stopped scaling. We help logistics teams move scheduling out of spreadsheets. One customer cut their weekly scheduling meeting from two hours to thirty minutes. Worth a 15-minute look next week? — Sam
Follow-up 1 (day 3): "Dana, one question that might be useful either way: who owns the schedule when a driver calls out?"
Tips and pitfalls
- Do the research yourself, then paste it in. AI can't reliably see a prospect's latest news, and it may make up details. Never send a "fact" about a prospect you haven't checked.
- Only use proof you can stand behind. Remove any numbers the AI added that weren't in your input.
- Keep it short. If the email needs scrolling on a phone, cut it.
- Don't paste confidential CRM data. into a tool your company hasn't approved.
- Follow email laws and your company's rules. (opt-outs, sender identity) for your region.
Works with
These prompts are plain text, so they work in ChatGPT, Claude, and Gemini. Paste each step into the same chat so the AI keeps the context from earlier steps. Free and paid tiers both work. Paid tiers usually handle longer pasted material better.
FAQ
Will prospects know it was written by AI?
They might if you send it unedited. Personal details and your own edits make it sound like you.
Can I do this for 50 prospects at once?
You can paste a list, but quality drops. Batches of 5-10 with real research work better.
What about LinkedIn messages?
Same steps. Just ask for under 300 characters in Step 3.
How do I handle replies and objections?
Paste the reply in and ask for 2 response options. Objection handling will get its own playbook.