Generate A/B Email Subject Lines

This prompt generates a ready-to-test set of six subject lines spread across three psychological angles — curiosity, benefit, and urgency — so you're comparing genuinely different approaches rather than six variations of the same idea. It also produces matching preheader text and a self-scored open-rate estimate with reasoning, which helps you prioritize what to actually send when you can only test two or three. Use it whenever you're about to send a campaign and want a structured A/B starting point instead of guessing.

What you can expect back

CURIOSITY
1. "We did the math on your plan" (7) — intrigue without revealing the discount.
   Preheader: The numbers might change how you renew this week.
2. "This won't be here Thursday" (6) — vague but creates a question.
   Preheader: A quick heads-up before the window closes.

BENEFIT
3. "Save 30% on your whole year" (8) — clear, concrete payoff up front.
   Preheader: Lock in annual pricing and forget about it.
4. "Your annual plan, now 30% less" (8) — direct and personally framed.
   Preheader: Same plan, lower price — for 48 hours only.

URGENCY
5. "48 hours: 30% off ends soon" (7) — explicit deadline drives action.
   Preheader: After Thursday, it's back to full price.
6. "Last call: 30% off your plan" (7) — scarcity language proven to lift opens.
   Preheader: Don't let the annual discount slip by.

Illustrative example — your results will vary by tool and inputs.

Get sharper results

• 01Give the model your actual sender name and audience (e.g. 'B2B SaaS admins, formal tone') so the variants match how your list expects to be spoken to.
• 02Only test two or three at a time — pick the highest-scored line from two different angles so your A/B reveals which psychology works, not just which wording.
• 03Tell it to avoid spam-trigger words and excessive punctuation if deliverability is a concern; ask it to flag any line likely to hit spam filters.
• 04Have it keep emoji out unless you ask — emoji rendering varies by client and can hurt as often as it helps for some audiences.
• 05Request a 'plain/control' seventh option with zero gimmick, so your test always includes a baseline to measure the clever ones against.

Adapt it for your case