Get 7 distinct subject lines for any email — instantly.
Describe the email — what it's about, who it's for. We generate 7 subject-line variants that stay in the 30-60 character sweet spot, avoid known spam triggers, and vary structure (question, statement, personalisation, number-led).
Subject lines are the single most A/B-tested feature in email, and the public data is remarkably consistent: short beats long (30-60 chars), specific beats generic (numbers, names, dates), and human-sounding beats marketing-sounding. The model is constrained to all three.
Spam triggers are heuristic — single words like 'free' aren't toxic in isolation, but in the wrong subject they tip filters that are already on the edge. The system prompt explicitly forbids the worst offenders ('act now', 'limited time', '100% free', 'click here', 'congratulations') so the output is safe to send from a clean mailbox.
Variety matters across batches. The prompt instructs the model to vary structure — question, statement, personalisation, number-led — so you can A/B-test directions rather than wording. Once you pick a winner, the email body matters more than further subject iteration.
Three is too few to A/B; ten clutters the comparison. Seven is the sweet spot we landed on after testing — enough variety, easy to scan at a glance.
The prompt allows up to one emoji per subject but doesn't insist on them. B2C emails sometimes lift open rates with one emoji; B2B almost never does. Add a "tone: casual" hint if you want emoji-friendly output.
Yes — paste any subject into our Subject Line Tester for an A-F grade across length, caps ratio, emoji count, punctuation, urgency, and spam triggers.
It works for any cold-or-warm B2B / B2C subject line. For newsletter-specific subjects ("Issue #42: …"), describe that in the context box and the model will follow.
The other half is whether the message arrives in the inbox at all. SieveGuard's Inbox Placement add-on sends test emails to seed accounts at every major ESP and reports inbox vs spam vs missing per provider.