A/B testing

Change one thing between version A and version B (a subject line, a send time, a button) and split the list at random. Pick the metric before you send, and let the test reach a sample large enough that the gap is not noise.

Testing several changes at once tells you that something worked, not what.

Two subject lines go to 2,000 random contacts each; the one with the higher open rate goes to the remaining list.

Invalid addresses bounce in both groups and skew the denominators. Verify and remove them first so the test measures reader behaviour, not list quality.

• Behavioral email
• Automation flow
• Verify a list
• Topic: Email marketing
• All terms