Every percentage point of conversion rate is revenue you already paid to acquire. That's what makes CRO the highest-leverage work available to most businesses — and also why it attracts so much nonsense. Button colours, urgency banners and best-practice checklists get sold as optimisation, but they're guesses dressed up as method, and a guess that happens to win teaches you nothing about the next one.
We work from evidence. Analytics tell us where people drop; session recordings and heatmaps show what they were trying to do; surveys and support tickets tell us what they couldn't find. That research produces hypotheses ranked by expected impact and effort, and each one states plainly what we expect to change and why. Then we test — properly, with the sample size calculated in advance, the duration set to cover full business cycles, and no peeking at partial results to declare an early winner.
Most tests don't win, and that's the job working correctly. A programme that reports 90% winners is measuring badly. What compounds is the learning: after six months you know what your customers respond to, which makes every future decision — pricing page, campaign, product launch — better informed than the last.