Picking what to test, in order of leverage

Picking what to test, in order of leverage
"Test button colors first." [BUSTED] It spread from a single green-vs-red story that became gospel. The truth: cosmetic tests need enormous traffic to detect tiny effects, so you waste months proving nothing. Prioritize by leverage:
— Rank changes by how deep in the page they sit. Headline and offer beat button color every time.
— Estimate the effect size honestly: messaging shifts move 10-30%, color shifts move maybe 1%.
— Filter by traffic: with under 1,000 conversions a month, only test changes big enough to see.
— Test one bold idea, not five timid tweaks competing for the same sample.
The nuance the color crowd skips: the biggest lever is what you're saying and to whom, not what shade you say it in. Test the offer before the pixels.