Good subject lines are specific, short and honest. The reliable way to find them is to generate a shortlist (AI is great at this), test two or three on a small hold-out, then send the winner to the rest. Treat opens as the metric for the subject and clicks as the metric for the email.
What makes a subject line work
- Relevance: it matches what is inside.
- Specificity: numbers, names, concrete words.
- Curiosity, but not click-bait. Promise something you deliver.
- Length: short enough to render on mobile.
- Sender name: matters as much as the subject itself.
How to test properly
Send each candidate to a small, random sample of your list, large enough to detect a meaningful open-rate difference. Then send the winner to the rest. See A practical guide to email A/B testing for sample-size guidance.
Where AI helps
AI generates twenty candidates in seconds, rates them against historical performance and proposes the best three to test. The human picks the tone; the model handles the brainstorm.
Things to stop doing
- ALL CAPS, excessive emojis and ten exclamation marks.
- False urgency (Last chance, when it isn't).
- Misleading personalisation (Re:, Fwd:).
Frequently asked questions
What is a good open rate?
It varies by industry and list. Compare to your own baseline first. Compare to the figures in [Email revenue benchmarks](/resources/guides/email-revenue-benchmarks) for context.
Should I use preview text?
Yes. Treat it as a second subject line. It often does more for opens than the subject itself.
Do emojis help?
Sometimes, in consumer lists, when they are on brand and used sparingly. Test, don't assume.