Definition
Variant testing (A/B)
Sending two versions of the same step to comparable prospects and comparing outcomes to learn which wording works.
Variant testing means writing two (or more) versions of one step in a sequence, usually differing in the subject line or the opening, and letting the campaign split prospects between them at random. After enough sends, you compare how each version performed and keep the better one. The randomisation is what makes the comparison fair: the two groups differ only in the wording they received.
It matters because opinions about copy are cheap and usually wrong, and a two-person team cannot afford to run a whole campaign on a hunch. Testing replaces the argument with a result. It also sets a trap: with small numbers, differences that look decisive are noise, and if you measure by opens you are measuring image loading rather than interest. Judge variants by replies, and above all by positive replies, and be patient enough to let a test run before you crown a winner. Change one thing at a time, or you will not know what made the difference.
In Seegnals each step in a sequence can have A/B variants of subject and body. The campaign stats include per-step statistics and a variant comparison table, and you can export the numbers as CSV to look at them your own way.