Case study: an A/B test that found the table position, not the button, was the win

Case study: an A/B test that found the table position, not the button, was the win

A review site assumed their low conversion was a weak CTA. They ran a structured test with an A/B tool on 40k visitors before touching button copy.

Test 1 — button color/copy: "Check Price" green vs "See Deal" orange. Result: +0.3%, inside noise. Nothing.

Test 2 — table above vs below the intro: moving the comparison table above the 400-word intro lifted affiliate clicks 19% at 95% confidence over 3 weeks.

The lesson: they nearly spent a month iterating button copy when the real lever was that readers wanted the verdict before the preamble.

Gotcha nobody mentions: their tool's default "95% confidence" stopped Test 1 early as a "loss" — but with only 0.3% delta it was underpowered. Always sanity-check sample size, not just the confidence badge.

My pick: test layout and content order before micro-copy — the big wins live in structure.

Bottom line: button color moved 0.3%, table position moved 19%. Test the skeleton before the paint.