ChatGPT vs Claude vs Gemini: Who Writes the Best Cold Email?
I Gave All Three AIs the Same Cold Email Brief — and the Gap Was Embarrassing
I sent the exact same brief to ChatGPT, Claude, and Gemini: write a cold email to a marketing director at a mid-sized e-commerce brand, pitching a freelance email copywriting service, aiming for a reply, not a sale. Same words. Same context. Three completely different emails came back — and one of them was so good I almost felt bad for the other two. If you're using AI to help you write cold emails (or thinking about starting), what I found will change which tool you reach for. The difference isn't just style — it's the number of replies you'll actually get.
The 3 Emails, Side by Side: What Each AI Actually Produced
The brief I used was specific on purpose: "Write a cold email from a freelance email copywriter to the marketing director of a mid-sized e-commerce brand. The goal is a reply, not a sale. Keep it under 150 words. Sound human, not salesy. Reference that email revenue is typically 30–40% of e-commerce revenue."
ChatGPT (GPT-4o) delivered something clean and competent. The subject line was "Quick question about your email revenue" — solid, not flashy. The body acknowledged the 30–40% stat, mentioned a recent win with a similar client, and ended with a soft yes/no question. It read like a decent SDR wrote it on a good day. The problem? It felt slightly templated. You could tell an AI touched it.
Gemini (1.5 Pro) went the opposite direction — too conversational, almost casual to the point of sounding unqualified. It opened with "Hey [Name], I've been following your brand and love what you're doing!" which is the cold email equivalent of showing up to a job interview in flip-flops. It also ignored the word count guideline and clocked in at 210 words. When your AI can't follow a brief, that's a red flag for the output quality.
Claude (Claude 3.5 Sonnet) did something neither of the others did: it wrote an email that felt like it came from a real person with a real opinion. The subject line was "Your welcome flow is leaving money on the table" — specific, slightly provocative, and impossible to ignore. The body was exactly 143 words, referenced a common gap in e-commerce email strategy without being generic, and ended with a question that made saying "yes" feel effortless. It was the only one I'd actually send.
Why Claude Won — and the Psychological Trick It Accidentally Used
Most people assume the best cold email is the most polished one. It's not. The best cold email is the most specific one — and Claude nailed specificity without being prompted to do so.
Claude's subject line — "Your welcome flow is leaving money on the table" — uses a technique copywriters call the implied insight. It tells the reader you already know something about their business before they've said a word. That creates immediate curiosity and a subtle power shift. The reader thinks: "How do they know that? Do I have a problem with my welcome flow?"
ChatGPT's subject line ("Quick question about your email revenue") is fine, but it's been used a thousand times. Inboxes are trained to ignore it. The implied insight approach gets opens because it feels personal even when it isn't.
Here's the deeper lesson: Claude tends to write with an editorial voice. It doesn't just complete the task — it makes micro-decisions about what would actually work. That's what separates a reply-worthy email from a deleted one. It's the difference between a tool that fills in words and one that thinks about persuasion.
You can push the other AIs to this level with the right prompt — but Claude gets closer on the first pass, which matters when you're writing 20 emails in a session.
How to Use This TODAY: A Cold Email Workflow That Actually Gets Replies
Start with Claude for the first draft. Use this exact prompt: "Write a cold email to [job title] at a [industry] company pitching [your service]. Goal: a reply, not a sale. Under 150 words. Sound like a confident human, not a pitch deck. Include one specific insight about a common problem in their industry that I can solve."
That last sentence — "include one specific insight" — is doing heavy lifting. It forces Claude to lead with value instead of leading with you, which is the single biggest mistake in cold outreach.
Once you have Claude's draft, run it through ChatGPT for a quick optimization pass. Use this prompt: "Here's a cold email I wrote. Rewrite the subject line 3 ways — one curiosity-driven, one direct, one provocative. Don't change the body." ChatGPT is genuinely excellent at rapid variation and subject line testing. Use it for that.
Then do one final check: paste the email into Claude again and ask: "Read this cold email and tell me what a skeptical marketing director would think in the first 5 seconds. What would make them delete it?" This reverse-engineering step takes 60 seconds and catches the lines that feel off before you hit send.
The whole workflow takes under 10 minutes per email. If you're sending personalized cold emails at volume — say, 20 a week — that's still under 4 hours for a week's worth of outreach that sounds human and lands differently in the inbox.
The Part Most People Get Wrong
Most people treat AI like a vending machine. They type "write me a cold email," get something generic, paste it into their CRM, and wonder why nobody replies. That's not an AI problem — that's a brief problem.
The output is only as good as the context you give it. If you don't tell the AI who you're writing to, what they care about, what their likely objection is, and what action you want — you'll get a forgettable email every single time, regardless of which tool you use.
The second mistake is trusting the first draft. Even Claude's impressive first draft isn't the one you should send. Use the draft as a skeleton, then edit one or two lines yourself — add a real detail, change a phrase that sounds slightly off, make the question at the end feel genuinely human. That 5% of human polish is what makes AI-assisted emails indistinguishable from the real thing.
The third mistake is using one AI for everything. ChatGPT, Claude, and Gemini have real strengths — and treating them as interchangeable is like using a screwdriver when you need a scalpel. Claude leads on persuasive first drafts. ChatGPT leads on rapid variation and structured tasks. Gemini is improving fast but isn't where it needs to be for nuanced copywriting yet. Use the right tool for the right job.
Key Takeaways
- Claude wins on cold email first drafts: Its editorial voice and specificity produce more reply-worthy emails out of the box than ChatGPT or Gemini.
- The implied insight subject line: Referencing a specific, plausible problem in the reader's business (before they've told you anything) dramatically increases open rates.
- Your brief is your bottleneck: Vague prompts produce vague emails — give the AI a job title, an industry, a goal, a word count, and a specific insight to lead with.
- Use ChatGPT for subject line variations: It's fast, creative under constraints, and great for generating 3–5 options to A/B test.
- The reverse-engineering check: Asking Claude "what would make a skeptical reader delete this?" is the fastest editing shortcut you're probably not using.
What to Do Right Now
Open Claude, paste this prompt, and fill in your details: "Write a cold email to a [job title] at a [industry] company pitching [your service]. Goal is a reply only. Under 150 words. Lead with one specific insight about a problem they likely have that I can solve. Sound confident and human." Run it, take 2 minutes to add one real personal detail, then use ChatGPT to generate three subject line options. You'll have a better cold email in the next 10 minutes than most people write in an hour.