A/B test
Send two variants of an email to comparable audiences and measure which performs better.
An A/B test (also called a split test) is the practice of sending two variants of an email — typically differing on one variable like subject line, sender name, or CTA copy — to two equivalent random samples of your list, then measuring which variant performed better on a chosen metric. The metric is usually open rate (for subject-line tests), click-through rate (for body and CTA tests), or conversion rate downstream of the click.
The math of email A/B testing is unforgiving. To detect a 1 percentage-point lift in open rate with statistical significance, you typically need 5,000+ recipients per variant. Smaller lists can still run tests, but the conclusions are directional, not statistical. Beware the temptation to declare a winner after 200 sends.
The strongest A/B tests vary exactly one variable. Multivariate tests (changing both subject and body) confound the signal. A/B winner-predictor tools like ours use Claude's rubric to predict winners on copy alone before you send, which is useful when your list is too small for statistical testing but you still want a defensible choice.