ChatGPT vs Claude vs Gemini: Rewriting a Bad Sales Email
We Fed the Same Cringe-Worthy Cold Email to All Three AIs โ Here's What Happened
Cold emails are killing deals before they even start. Most salespeople know their outreach is bad, but they don't know how bad โ or how to fix it fast. We took one genuinely terrible cold email (the kind that makes you wince when you read it) and fed it to ChatGPT, Claude, and Gemini with the exact same prompt. What came back surprised us. One rewrite was so good we'd actually send it. One was polished but still missed the point. And one made the original look professional by comparison. Here's the full breakdown โ with the actual outputs โ so you know exactly which tool to reach for next time your pipeline is suffering.
The Original Email Was Bad in All the Ways That Matter
Before we get to the rewrites, you need to see the source material. This is the cold email we used โ pulled from a real submission in a sales community (shared anonymously):
"Hi [First Name], I hope this email finds you well. My name is Jake and I am reaching out because I believe our solution could really benefit your company. We offer a wide range of services that help businesses like yours achieve their goals. I would love to schedule a 30-minute call to discuss further. Please let me know your availability. Best regards, Jake."
This email has every sin in the book. No specific pain point. No personalization. A vague offer. A commitment-heavy ask. "I hope this email finds you well" alone is enough to get deleted.
The prompt we gave all three AIs was identical: "Rewrite this cold sales email to make it more compelling, specific, and likely to get a reply. Keep it under 150 words. The sender sells project management software to marketing agencies."
We didn't give extra context. We didn't prompt-engineer around weaknesses. We wanted to see what each AI does with the same raw input โ because that's how most people actually use these tools.
Same email. Same prompt. Three very different results.
ChatGPT's Rewrite Was Clean โ But It Played It Safe
ChatGPT produced a solid rewrite. Fast, readable, and clearly better than the original. Here's what it gave us:
"Hi [First Name], Marketing agencies managing 10+ client projects usually hit the same wall โ too many tools, too many status meetings, too much time lost. That's exactly why we built [Product] โ one workspace where your team tracks every campaign, deadline, and deliverable without the chaos. Agencies like [Similar Agency] cut their project admin time by 40% in the first month. Would it make sense to show you how it works for a team your size? Happy to keep it to 15 minutes."
That's a good email. Specific pain point. Social proof. Reduced ask (15 minutes, not 30). ChatGPT understood the assignment.
But here's the catch: it's almost great. The opening line is strong, but it still leans on a generic agency archetype rather than anything truly tailored. The social proof is fabricated with a placeholder, which means you're doing the work of filling it in anyway.
ChatGPT's output is what you'd call "competent first draft" energy. It gets you 80% of the way there quickly. If you're sending volume outreach and need a reliable baseline, this is your tool.
The real weakness? It didn't push back on anything or suggest a better approach. It just did what you asked. Sometimes that's exactly right. But for sales emails specifically, there's more value in a tool that thinks about strategy, not just structure.
Claude Did Something None of the Others Did โ It Questioned the Brief
Claude's rewrite was good. But what made it stand out wasn't the email itself โ it was what came with the email.
Before delivering the rewrite, Claude flagged two things: the placeholder social proof would only work if real, and it offered two alternative versions โ one direct response style and one curiosity-based opener. No other AI did this unprompted.
Here's Claude's primary rewrite:
"Hi [First Name], Most marketing agencies I talk to are running projects across three or four different tools โ and still dropping balls. [Product] pulls everything into one place: client timelines, team tasks, approvals, reporting. Setup takes a day. The ROI shows up in week one. I won't pitch you on a long call โ a 12-minute screen share is usually all it takes to know if this fits. Worth a look?"
That closing line โ "Worth a look?" โ is doing heavy lifting. It's low commitment, conversational, and disarming. It doesn't beg. It invites.
Claude's ability to offer strategic options rather than a single output is what separates it here. It understood that the best rewrite depends on the audience, and it gave you tools to make that decision yourself.
If you're writing high-stakes, high-personalization outreach โ think enterprise deals, warm leads, or anything where one email could open a six-figure door โ Claude is the tool you want. It thinks more like a strategist than a copywriter.
Gemini's Rewrite Taught Us What "More Words" Doesn't Mean
Gemini's output was the longest. It was also the weakest.
Here's a condensed version of what Gemini produced:
"Hello [First Name], I hope you're having a great week! I came across your agency online and was really impressed with the work you're doing. As someone who works in the project management software space, I thought it would be worthwhile to connect. Our platform is designed to help marketing agencies like yours streamline workflows, improve team collaboration, and deliver projects on time and within budget..."
It kept going. By the time it reached the call to action, it had already lost the reader twice. The opener is nearly identical to the original bad email. "I hope you're having a great week" is just a minor upgrade from "I hope this email finds you well."
Gemini added length but not substance. It leaned on buzzwords โ "streamline workflows," "improve collaboration" โ without replacing them with specifics. This is the classic AI failure mode: pattern-matching to what a sales email looks like instead of what makes one work.
To be fair to Gemini, it performs better when you give it more structure upfront. If you add "Avoid generic openers, avoid buzzwords, focus on one specific pain point" to your prompt, the output improves meaningfully. But the fact that you need to do that extra work โ and ChatGPT and Claude didn't require it โ matters.
For cold email rewriting specifically, Gemini in its default state is not your tool right now.
The Part Most People Get Wrong
Most people treat AI like a vending machine. You put in a bad email, you expect a perfect email to come out. That's not how this works โ and it's why so many AI-assisted emails still sound like AI-assisted emails.
The prompt is the product. The email is just the output. If you give a vague prompt, you get a vague rewrite. "Make this better" is not a prompt. "Rewrite this cold email for a marketing agency decision-maker who's skeptical of new software tools. Use a conversational tone, open with a pain point, include one specific result, and end with a low-commitment ask under 10 words" โ that's a prompt.
Most people also pick one AI and stick with it religiously. That's leaving results on the table. The real workflow is to use ChatGPT for a fast first draft, Claude for a strategic second pass or when the stakes are high, and skip Gemini for copywriting tasks until it catches up. Different tools have different strengths โ treating them as interchangeable is a mistake.
Finally, the biggest error: not editing the output. AI gives you a strong starting block, not a finished race. The social proof placeholder, the company name, the specific agency reference โ those gaps are where you create the actual personalization that closes deals. AI writes the skeleton. You add the muscle.
Key Takeaways
- ChatGPT: Fastest, most consistent rewriter โ ideal for high-volume outreach where you need a reliable 80% solution quickly.
- Claude: Best for high-stakes emails โ it thinks strategically, offers options, and challenges weak briefs instead of just executing them.
- Gemini: Currently the weakest for cold email tasks โ requires heavy prompt engineering to match what the others do out of the box.
- Your prompt determines your output: Vague prompts produce vague emails โ always specify tone, audience, pain point, and call-to-action style.
- AI is a drafting partner, not a ghostwriter: The personalization, real social proof, and final polish still have to come from you.
What to Do Right Now
Open Claude right now, paste your worst-performing cold email, and use this prompt: "Rewrite this cold sales email. I sell [your product] to [your audience]. Avoid generic openers and buzzwords. Open with a specific pain point, include one credible result, and close with a low-commitment ask under 10 words. Give me two versions." You'll have two better emails in under two minutes โ and you'll immediately see which AI approach fits your voice best.