Scientific testing, controls, and what to actually A/B (and what not to).
Every ad, email, and landing page has a control, the current winner. Challengers are the hopeful replacements. Operating the control/challenger system correctly is what turns testing into a compounding engine.
Testing is only as good as the measurement. Track the right numbers with the right attribution and testing compounds. Track the wrong ones and you optimize toward noise.
systematized advertising testing in 1923. A century later, most marketers still guess instead of test. The operators who do test build compounding knowledge that competitors can't match.
A complete checklist of testable elements, ordered by likely impact. Most operators test the wrong things. Here's the right order.