I Sent the Same Cold Email Brief to ChatGPT, Claude, and Gemini — The Results Exposed a Gap Most Marketers Don't Know Exists
Most people assume all AI writing tools produce roughly the same output. They pick one, use it forever, and never question what they're leaving on the table. I ran a head-to-head test — same brief, same context, three different AIs — and the differences weren't subtle. We're talking about the gap between a cold email someone ignores and one that actually books a call. By the end of this article, you'll know exactly which AI writes the best cold emails, when to use each one, and how to squeeze better output from all three.
The Test: Same Brief, Three AIs, Zero Mercy
I gave all three AIs the exact same prompt — no tweaking, no favoritism. The brief was:
"Write a cold email from a freelance UX designer named Alex to the Head of Product at a mid-sized SaaS company. The goal is to book a 15-minute call. Alex has redesigned onboarding flows that reduced churn by 23% for a previous client. Keep it under 150 words. Tone: confident but not pushy."
Simple enough brief. Clear goal. Measurable constraint. The kind of prompt a real freelancer or agency would actually use.
ChatGPT (GPT-4o) delivered fast and clean. The subject line was sharp — "Cut churn like we did for [Client] — 15 min?" — and the body led immediately with the 23% stat. It felt like a real email a confident person would send. The CTA was specific: "Are you free for a quick call Thursday or Friday?"
Claude (Claude 3.5 Sonnet) took a more conversational angle. Less punchy, more human. Instead of leading with the stat, it opened with a single observation about common onboarding friction, then dropped the 23% figure two sentences in as proof. The tone felt warmer — almost like a message from someone you'd already half-met at a conference.
Gemini (Gemini 1.5 Pro) gave the longest response and, honestly, the weakest cold email. It was polite, well-structured, and thoroughly average. The subject line was "Exploring UX Collaboration Opportunities" — which is the kind of subject line that gets buried under three other unread emails before lunch.
Why Claude's "Warm Logic" Approach Actually Outperforms in Most Inboxes
Here's what most AI comparison articles miss: cold email success isn't just about being punchy. It's about passing the "real person" filter that every busy professional runs in their head when they scan their inbox.
Claude has been trained with an unusually strong emphasis on natural, coherent conversation. That shows up in cold emails as something I call "warm logic" — where the email feels reasoned and human, not assembled from a template. It doesn't just tell you the writer is good. It makes you feel like you're talking to someone who actually gets your problem.
In the test email, Claude wrote: "Most SaaS onboarding drops users at the same two friction points — and most teams don't catch it until the churn numbers land. That's exactly what I fixed for [Client], cutting their churn by 23%." That's not just confident. That's the kind of framing that makes a Head of Product stop and think, "How does this person know that?"
The mental model here is "diagnosis before prescription." Claude naturally structures its writing to identify a problem before offering a solution. For cold emails — where your reader is skeptical and skimming — this approach earns micro-credibility in the first sentence. ChatGPT tends to skip straight to the pitch. Claude earns the right to pitch.
Where Claude loses points: it sometimes runs slightly long and can soften a CTA when you need it to be direct. A quick follow-up prompt like "Make the call-to-action more direct and time-specific" fixes this in seconds.
How to Use All Three AIs Together for a Cold Email That Converts
You don't have to pick one and pray. The most effective workflow I've found uses all three AIs as a drafting-editing-sharpening pipeline — and it takes under 15 minutes.
Step 1: Start with Claude for your first draft. Use the same brief structure from above — name, role, specific result, word count, tone. Claude's warm logic gives you a solid human-feeling foundation that won't set off anyone's "AI email" radar.
Step 2: Take Claude's draft into ChatGPT and sharpen the opening line and subject. Use this prompt: "Here's a cold email draft. Rewrite just the subject line and first sentence to be more direct and curiosity-driven. Don't change anything else." ChatGPT is excellent at punchy, high-conversion hooks. Let it do what it does best.
Step 3: Paste the revised email into Gemini and ask it to stress-test. Use: "Read this cold email as a skeptical Head of Product who gets 50 emails a day. What would make you delete it immediately? What would make you reply?" Gemini is actually strong at analytical critique even when its generative writing is average. It'll catch vague claims, weak CTAs, or anything that sounds too salesy.
Step 4: Make your final edits yourself. AI can get you to 85%. The last 15% — personalizing the company reference, tightening one specific line, making the CTA match your actual calendar — that's your job. It takes two minutes and it's the difference between a template and an email that feels written for that one person.
You can run this entire pipeline in 15 minutes. If you're sending cold emails at volume, batch this — run 10 briefs through Claude at once, then move to ChatGPT, then Gemini.
The Part Most People Get Wrong
Most people treat AI cold email prompts like a vending machine — drop in a basic request, get out something usable. That's wrong, and it's costing you replies.
The single biggest mistake is not giving the AI your result before your role. Most prompts start with "I'm a UX designer and I want to reach out to..." — which primes the AI to write a generic intro. Instead, lead with the outcome: "I reduced onboarding churn by 23%" should be the first thing in your brief. The AI anchors on whatever comes first. Give it something worth anchoring on.
The second mistake is accepting the first output as final. Every AI — including Claude at its best — writes a first draft. The best cold emails come from iterating with follow-up prompts like "Make the opening line 10 words or fewer" or "Remove any sentence that sounds like it came from a template." Two rounds of iteration takes three minutes and the quality difference is significant.
Finally, people forget to specify the reader's role and pain, not just their own pitch. The more you tell the AI about who is reading the email — their job title, their pressures, what they probably hate about cold emails — the more targeted the output becomes. A prompt that includes "Write this for a Head of Product who is under pressure to reduce churn this quarter" will outperform "Write a cold email to a Head of Product" every single time.
Key Takeaways
- Claude wins for tone: Its "warm logic" structure makes cold emails feel human and earns credibility before the pitch lands.
- ChatGPT wins for hooks: It writes sharper subject lines and opening sentences — use it to punch up any draft.
- Gemini wins for critique: It's a stronger analytical editor than a generative writer — use it to stress-test, not draft.
- The pipeline beats any single tool: Claude → ChatGPT → Gemini used in sequence produces better results than any AI used alone.
- Your brief is everything: Lead with your result, specify your reader's pain, and iterate — the prompt quality determines the output quality more than which AI you use.
What to Do Right Now
Open Claude right now and paste this exact brief — swap in your own details: "Write a cold email from [your name/role] to [target job title] at a [company type]. Goal: book a 15-minute call. My key result: [specific number or outcome]. Under 120 words. Tone: confident, direct, not salesy." Run the output through ChatGPT to sharpen the subject line, then send it to your next five prospects. You'll have a better cold email than 90% of what's hitting inboxes today — and you'll have built it in under 15 minutes.