Write email subject lines with ChatGPT

What you'll make

A set of email subject lines in 3 different styles, plus an A/B testing plan so you know which one actually gets the highest open rate. Instead of spending 20 minutes staring at one subject line, you'll have options to test.

What you need

• A ChatGPT account
• The email you're sending (or at least its topic and purpose)
• An email tool that supports A/B testing (Mailchimp, Klaviyo, ConvertKit, most do)
• 15 minutes

Steps

1. Tell ChatGPT about the email and ask for subject lines:

I'm sending an email to my list about [topic — e.g. "a new product launch", "a weekly newsletter", "a flash sale on winter items"]. The email's main purpose is [get them to click through / get them to reply / get them to read an article].

My audience is [who they are — e.g. "Etsy sellers", "people who bought from my store before"].

Generate 10 subject lines in 3 different styles:

Direct (3 options): says exactly what's in the email. No mystery, just clarity.
Curious (4 options): creates intrigue without being clickbait. The reader wants to open to find out more.
Urgent (3 options): implies time-sensitivity without using spam-trigger words. Not "URGENT" or "ACT NOW" — more like "tonight's the last night" or "closing in a few hours".

For each subject line:
1. The text (under 50 characters)
2. Which style it is
3. Why it works for this email
4. Preview text to pair with it (under 90 characters — the snippet that shows after the subject in the inbox)

Avoid these words: free, urgent, act now, limited time, last chance, don't miss out, hurry.

2. Pick 2 subject lines from different styles to A/B test. Ask ChatGPT for the testing plan:

I want to A/B test these 2 subject lines:
A: [subject line 1]
B: [subject line 2]

Give me a simple testing plan:
1. What percentage of my list should get each variant?
2. How long should I wait before sending the winner to the rest?
3. What metric determines the winner (open rate vs click rate)?
4. Is my list big enough for the result to be meaningful? (My list has [number] subscribers.)
5. What should I do with the winner — just send it, or save the pattern for next time?

Keep it practical for a small list — not enterprise-level advice.

3. Ask ChatGPT to track patterns over time:

I'll be testing subject lines regularly. Give me a simple system to track what works:
1. What to record after each send (subject line, style, open rate, click rate, list size)
2. How to spot patterns after 5–10 sends (e.g. "curious subject lines consistently outperform direct ones for my list")
3. When to stop testing and just use what works
4. A simple format for tracking (a table I can copy into a spreadsheet)

Keep it low-effort — I don't want to spend more time tracking than writing.

4. Set up the A/B test in your email tool. Send variant A to half the list and variant B to the other half (or use your tool's built-in A/B split feature). After the waiting period, send the winner to the remaining subscribers.

5. Record the results in your tracking spreadsheet. After 5–10 emails, look for patterns in which style works best for your audience.

Check your result

• You have 10 subject lines across 3 styles, each under 50 characters
• The 2 you're testing are from different styles (not 2 direct lines)
• No subject line uses spam-trigger words
• Each subject line has paired preview text that complements it (not repeating the same words)
• You have a simple tracking system that takes under 2 minutes per send

If your list is under 100 subscribers, A/B test results won't be statistically significant — a 2% difference in open rate could be pure chance. That's fine. You're looking for patterns over many sends, not calling a winner from a single test. Send the subject line you think is best to the full list, and start tracking patterns once you have more data.