The short answer
A good ChatGPT cold email prompt gives the model three things it can't guess: your product context, real research about the prospect pasted in, and strict rules for length, tone and structure. Ask it to use only the facts you provide, stay under 120 words and end with one simple question. Then edit the draft into your own voice before sending.
On this page
- Why Plain ChatGPT Cold Emails Sound Generic
- The 3 Inputs Every ChatGPT Cold Email Prompt Needs
- 6 Copy-Paste ChatGPT Cold Email Prompts
- Where to Get the Research to Paste In
- What to Check Before You Send a ChatGPT Draft
- When a Purpose-Built Tool Saves the Hour
- Skip the prompting and get researched cold emails in three steps
- FAQ
Key takeaways
Context in, specifics out: ChatGPT only knows what you give it. Without your offer and real research, it fills the gaps with generic sales language.
Paste the research: Copy in their homepage, a job post or a news item, and tell the model to use only those facts.
Set hard rules: Word limit, banned phrases, one question at the end. Constraints are what make the output sound human.
Verify every fact: Models can state wrong details with confidence. Check every claim about the prospect before you send.
Why Plain ChatGPT Cold Emails Sound Generic
ChatGPT writes the most likely email for the input it gets. Give it a one-line request and the most likely email is the one every other rep is sending, which is exactly why buyers can spot it.
| What ChatGPT gets | What it writes | The fix |
|---|---|---|
| No product context | "We help companies like yours drive growth." | Paste a short product brief |
| No prospect research | "I was impressed by your innovative approach." | Paste real facts about the account |
| No rules | 200 words, three questions, a P.S. | Set a word limit and one ask |
| No voice sample | Polished corporate tone | Paste an email you wrote that worked |
| No real proof | "Our clients see significant results." | Give it one true customer outcome |
The model isn't the problem; the input is. Every fix in the right column is something only you can provide. Once you do, the drafts get noticeably sharper.
The 3 Inputs Every ChatGPT Cold Email Prompt Needs
Every prompt below draws on the same three inputs. Prepare them once and the rest is copy and paste.
- Your product context. What you sell, who it's for, the problems it solves, one or two proof points and phrases to avoid. Write it once and reuse it.
- Prospect research. Real text from their website, a job post, a news item or the contact's public post. Paste it in rather than asking the model to remember or guess.
- Your rules. Length, structure, tone and banned phrases. These do more for quality than any clever wording in the request.
Save the context somewhere reusable. Custom instructions or a ChatGPT Project keep your product brief available, so each new chat starts from the same baseline.
6 Copy-Paste ChatGPT Cold Email Prompts
Replace every [bracketed] part before you run a prompt. Run them in order in one chat: context first, then research, then the email.
Prompt 1: Set your product context
You are helping me write B2B cold emails. Here is my context.
Use it for every email in this chat.
Product: [one-line description]
Who it's for: [roles and company types]
Problems it solves: [two or three, in plain words]
Proof: [one or two real customer outcomes I can share]
Tone: [e.g. direct, plain, peer-to-peer]
Never say: [e.g. "hope this finds you well", "game-changing", "synergy"]
Reply "ready" and wait for the prospect details.
Prompt 2: Summarize the prospect research
Below is text from [Company]'s website, careers page and news.
List the three facts most relevant to the problems my product solves.
For each fact, quote the source text and say why it might matter.
Use only the text below. If nothing is relevant, say so.
[paste research text]
Prompt 3: Write the cold email
Write a cold email to [Name], [title] at [Company].
Use only the facts from your summary. Do not invent details.
Structure:
1. One sentence on the most relevant fact and what it likely means for them.
2. One sentence on the problem it suggests.
3. One sentence of proof from my context.
4. One yes-or-no question as the ask.
Under 120 words. No greeting fluff, no exclamation marks, no P.S.
Prompt 4: Write subject lines
Write 5 subject lines for the email above.
Each must be 2 to 6 words, specific to [Company], and accurately
reflect the email's content. Use a question mark in two at most.
No clickbait, no "quick question", no fake "Re:".
Prompt 5: Critique and de-AI the draft
Act as a skeptical [prospect's title] who gets 50 sales emails a week.
Point out every phrase in the email that sounds generic, salesy
or AI-written, and any claim that seems unsupported.
Then rewrite it to fix those issues, keeping it under 100 words.
Prompt 6: Write follow-ups with a purpose
Write two follow-up emails for this prospect, to send as replies
in the same thread on day 3 and day 7.
Follow-up 1 adds one new piece of proof or detail.
Follow-up 2 asks if someone else owns this problem.
Each under 60 words. Never write "just following up" or "bumping this".
For more prompts beyond cold email, our sales prompts library covers research, calls and objections, and the cold email subject line generator guide has 45 subject lines to compare against.
Where to Get the Research to Paste In
The research input matters more than the prompt. Spend five minutes collecting real text, and the model has something specific to work with. Our guide to how to research a prospect covers the longer routine.
- Homepage and About page. How they describe their customers and product, in their own words.
- Careers page. One or two job posts related to your offer. Hiring is often the best signal.
- Newsroom or blog. Launches, expansions and funding from the last quarter.
- The contact's public posts. Copy the text of a relevant post rather than asking the model to find it.
If you use ChatGPT's web search, verify everything. Ask for a source link per fact and open each one. Models can confuse companies with similar names or present old news as current.
What to Check Before You Send a ChatGPT Draft
ChatGPT gets you to a decent draft fast. The last mile is yours, and it's where many "AI-sounding" emails could have been saved.
- Every fact is true. Check each claim about the prospect against its source.
- It sounds like you. Read it aloud. Rewrite any sentence you wouldn't say on a call.
- One ask only. Delete extra questions and any second call to action.
- No stock phrases slipped in. Scan for "I hope", "I wanted to reach out" and "innovative".
Our AI cold email writer guide has a fuller list of AI red flags and a quick editing checklist.
When a Purpose-Built Tool Saves the Hour
The prompts work, but the workflow doesn't scale well. For each account you collect research, paste it, run four prompts, check facts and edit, which is easily 15 to 25 minutes before you've sent anything.
| Step | ChatGPT workflow | Purpose-built tool |
|---|---|---|
| Product context | Paste or keep in a Project | Stored once in a playbook |
| Prospect research | Collect and paste by hand | Read from the prospect's website |
| Sources | Ask for them, then check | Listed beside each fact |
| Angles | One prompt per variant | Several variants per run |
| Follow-ups | Separate prompt | Included with day offsets |
Use ChatGPT for one-offs, a tool for the list. For a single important email, the prompts above are plenty. For a week of prospecting, the Edithly cold email generator reads each prospect's website and writes several angles with follow-ups and the sources behind every fact. For other ways to use ChatGPT across the sales job, see ChatGPT for sales.
How Edithly does it
Skip the prompting and get researched cold emails in three steps
Edithly does the research step for you: it reads the prospect's website and public signals, writes several email angles from your playbook and lists every fact it used with a source, so there's nothing to paste in.
- 1
Click the Email Outreach tile
Your product context lives in your playbook, read from your website once, so you never paste it into a prompt again.
- 2
Paste the prospect's website
Enter the company name and website, plus optional contact details, notes and a tone such as 'direct, no hype'.
- 3
Generate, check sources, copy
Click Generate email outreach. Review the personalization basis and do-not-say list, then copy the email and follow-ups into your inbox or sequencer.
Website
Ready
- Variant A · Hiring angle
- Subject lines
- Body
- Follow-ups · Day +3, +7
- LinkedIn note
- Do not say
- By hand
- 15 to 25 minutes per account researching, prompting and editing
- With Edithly
- A couple of minutes per account, research included
Frequently asked questions
What is the best ChatGPT prompt for a cold email?
One that includes your product context, pasted research about the prospect and strict rules. For example: 'Using only the facts below, write a cold email under 120 words to [name], [title] at [company]. Open with one specific fact, name the likely problem, add one line of proof and end with a yes-or-no question.' Then paste the facts.
Why do ChatGPT cold emails sound generic?
Without specific input, the model writes the most likely sales email, which is close to the average of every sales email it has seen. That means stock phrases, vague value claims and flattering openers. Feeding it real research, your own proof and a list of banned phrases pushes it away from the average.
Can ChatGPT research a prospect for a cold email?
ChatGPT can search the web when that feature is on, but it may mix up companies, use outdated information or fill gaps with plausible guesses. Ask it to cite a source for each fact and to say 'not found' instead of guessing, then open the links yourself before using anything in an email.
Is it okay to send cold emails written by ChatGPT?
Yes, as long as you check and own every word. Verify each fact about the prospect, rewrite anything that doesn't sound like you and make sure claims about your product are true. Avoid pasting confidential customer or company data into prompts unless your account settings and company policy allow it.
How is Edithly different from using ChatGPT for cold emails?
Edithly is built for this one job. It stores your product context in a playbook, reads each prospect's website when it generates, writes several email angles with follow-ups and a LinkedIn note, and lists the facts it used with sources. ChatGPT is more flexible; Edithly removes the research and prompting steps for each account.