A/B, randomized, ablation, or controlled intervention

Change one factor while a comparable group stays unchanged, creating the most credible version of 'what would have happened otherwise.'

Open rates jumped the week you tried the new subject-line style, and you're ready to declare victory. Except you also moved your send time that same week, and a competitor's newsletter went on break. Three things changed; the credit is up for grabs, and your memory will hand it to your favorite.

The move is to change one factor while a comparable group stays unchanged, which creates the most credible version of "what would have happened otherwise." The cue is wanting to know whether a change caused a result, not just preceded it. Steps: randomly split your list in half, send the two subject-line styles at the same time with everything else identical, pre-decide the metric that counts (opens, or better, clicks), then read the difference between groups. Because the halves were randomized, whatever else is going on that week hits both equally, so the gap belongs to the subject line. When it works, an opinion war inside your head becomes a number. Don't use it when randomizing would be unethical or unkind to the people involved, or when your sample is too small to detect a difference big enough to matter.