Run link experiments with ascsubtag or you're optimizing on vibes

Without split tracking, 'this CTA works better' is a guess. The experiment framework:
— Pick ONE variable: button text, link position, table vs inline
— Tag each variant with a distinct ascsubtag (e.g. cta-buy vs cta-check)
— Run until each variant has ≥300 clicks — under that, noise dominates
— Compare conversion AND EPC, not clicks alone (more clicks ≠ more orders)
— Hold everything else constant; one change per test
— Ship the winner site-wide, then test the next variable
A 'Check Price on Amazon' CTA beat 'Buy Now' by ~17% on conversion in one such test (n=900) — readers resist 'buy,' accept 'check.'
So what: one disciplined ascsubtag test per month beats a year of guessing which tweak earns.