Playbook

Cold email A/B testing guide (what to test and how)

Test one variable at a time — subject, first line, CTA, or send time — measure on reply rate (not opens), and give each test enough volume to be real. Here's a disciplined A/B testing approach.

25 Apr 2026 6 min readBy Autocloz Editorial, Deliverability team
Cold email A/B testing guide (what to test and how)

Short answer: A/B test cold email by changing one variable at a time (subject line, first line, CTA, or send time), measuring on reply / positive-reply rate rather than opens (which Apple MPP inflates), and giving each variant enough volume to be statistically meaningful before you call a winner. Test the highest-leverage element first — usually the first line and the CTA, not the subject.

What to test (in order of impact)

  1. First line / hook — relevance drives replies most.
  2. CTA — low-friction ask vs direct ask.
  3. Subject line — short/specific vs curiosity.
  4. Send time / day.
  5. Length — shorter usually wins.

How to test properly

  • One variable per test — or you can't attribute the result.
  • Measure replies, not opens.
  • Enough volume — a 20-email test proves nothing; aim for hundreds.
  • Kill the loser, scale the winner, then test the next variable.

Autocloz A/B-tests variants per step and reports reply + positive-reply rate, so you optimise on the metric that books meetings.

> Start free — A/B test on replies and compound the wins.

Share
Free to start

Stop reading. Start sending.

Every tactic in this article is implemented behind the Autocloz dashboard.