How to A/B test cold emails
A/B testing turns guesswork into learning, but only if the test is clean: one variable, enough volume, the right metric.
Small, consistent tests compound into a much stronger playbook.
Short answer
To A/B test cold emails, change one variable at a time (subject line, opener, CTA, or send time), split a large enough sample, measure the metric that matches the variable (opens for subject lines, replies for body), and keep winners in a library. Disciplined testing compounds into steadily better outreach.
Step by step
Test one variable at a time
Change only the subject line, the opener, the CTA, or the send time. Changing several at once makes the result uninterpretable.
Use a meaningful sample
Split enough sends that the result is not noise. Tiny tests produce random winners you cannot trust.
Match the metric to the variable
Judge subject lines by open rate and body or CTA changes by reply rate. Using the wrong metric hides the real effect.
Keep the winner
Promote the winning variant and retire the loser. Save winners in a library sorted by scenario.
Keep testing
Always have a test running. Continuous, disciplined testing is how outreach improves over time.
How Ardovo runs this for you
Ardovo is an AI-native CRM, so this is not just theory. Rook can build the sequence, draft each touch from your real contact and deal data, choose send times, and run the cadence across email and other channels while logging every reply against the record. You set the strategy in this guide; Rook executes it and flags the prospects worth your attention.
Frequently asked questions
How do I A/B test cold emails?
Change one variable at a time (subject line, opener, CTA, or send time), split a large enough sample, measure the metric that matches the variable, and keep the winner. Clean, disciplined tests produce trustworthy learnings.
What should I test in a cold email?
Subject lines (judged by opens), the opener and CTA (judged by replies), the length, the angle, and the send time. Test the highest-leverage elements first, one at a time, and build a library of winners.
How big should my A/B test sample be?
Large enough that the difference is not random noise, which depends on your volume and the size of the effect. Small tests produce misleading winners, so give each variant a meaningful number of sends before deciding.