The most common A/B test is subject line: split a sample (5–50% of the audience) between Variant A and Variant B, watch a chosen metric (open rate or click-through rate) for a set duration (1–48 hours), then send the winning subject to the remaining contacts. Some platforms automate the winner selection; others require manual review.
Body, send time, sender name, and CTA design are also testable, but only one variable at a time. Multi-variable tests are mathematically valid only if the audience is large enough to power each combination — usually tens of thousands of recipients.
A/B testing only works at sufficient scale. With a list of 500 contacts, the difference between two subject lines is almost always within statistical noise. Focus on quality (segmentation, content) below ~5,000 recipients; introduce A/B testing as part of the discipline above that threshold.