How do you approach A/B testing, and what would you test first?

I test one variable at a time, on a segment large enough that the result is not noise. With a small list, a difference of a couple of percentage points is meaningless, and acting on it makes the programme worse.

I start with subject line and sender name, because they gate everything downstream and are cheap to change. Then the call to action — wording, placement, and whether it is a button or a link. Then content structure: length, single offer versus multiple, plain text versus designed HTML. Landing page and form length come next, since the email's win is wasted if the page does not convert.

The metric matters: optimising subject lines against open rate alone reliably produces clickbait that opens well and converts badly. I judge on the downstream action — clicks, and where possible form submissions.

And I record results somewhere durable. A test that nobody remembers gets re-run in a year.