After 480 tests across 60 e-commerce stores, the pattern is humbling: most tests do nothing. The winners cluster around five levers — and almost none of them are 'make the button bigger'.
One: speed. Every 100ms of load time costs roughly 1% of conversion. Two: checkout friction. Not the number of fields, but the number of decisions. Three: social proof that arrives before the first scroll. Four: product-page clarity — one hero message, not three. Five: post-purchase experience, which quietly drives repeat revenue.
Our most surprising win came from a test we almost cut: removing the promotional banner entirely. Revenue per session rose 14%. The loudest element on the page was stealing attention from the reason people came.
CRO is a discipline of subtraction. The page that converts best is usually the one that said no to more things — more banners, more popups, more competing asks. Test that instead of the button color.
Diagnose before you optimise. Most conversion problems are not psychology problems, they are analytics problems: a leaky funnel step nobody instrumented, a payment method two thirds of your market prefers and you do not offer, or a search that returns nothing for the exact product name. Three hours of funnel analysis beats three months of button-colour testing.
The checkout is where CRO goes to retire. Guest checkout, one-page flows, address autocomplete, wallets, and honest delivery dates on the product page move more revenue than any headline ever written. Remove one form field and watch the needle. Then remove another.
Run experiments with discipline. A hypothesis, a minimum sample size, one variable, and the patience not to peek. Sequential testing on a small store is mostly theatre; you need either traffic or a very pointed change. When in doubt, ship the obvious fix and spend your statistics budget where the uncertainty actually is.
Speed converts. Every additional second of load time on mobile costs a measurable slice of revenue, and shoppers read site speed as a trust signal whether they know it or not. Compress hero media, defer what is below the fold, and measure revenue per session, not just page views.
Retention is the cheapest CRO available. Post-purchase flows, replenishment reminders and a loyalty mechanic that actually rewards spend will outperform your tenth landing-page test. The needle moves furthest for the store that treats the second purchase as the first campaign.
Enjoyed this? Let's talk about applying it to your product.
