Most redesigns change how a website looks and nothing about what it earns. The layout gets cleaner, the type gets bigger, the launch gets celebrated — and three months later the conversion rate is exactly where it started. A redesign only pays for itself when the decisions behind it are tied to what visitors are actually trying to do.
Across 650+ delivered projects we keep seeing the same handful of patterns move the number, and the same handful of “best practices” do nothing. This guide covers both, with the order we apply them in.
Article key takeaways
- A redesign that only changes appearance rarely changes revenue — diagnose before you design.
- The hero carries the outcome, one primary action and a proof marker in the first viewport.
- Structural fixes (speed, clarity, friction) outperform visual polish on every audit we run.
- Measure against a single primary conversion, not a dashboard of vanity metrics.
Why the first 600px decide the sale
Visitors decide whether to keep reading before they scroll. That means the hero has one job: state what you do, who it’s for, and what proof backs it — in language a customer would use, not a tagline the founder likes. Every hero we ship carries a specific outcome, one primary action, and a credibility marker close enough to be read in the same glance.
Helio hero — outcome-led headline, single primary action, proof strip within the first viewport.
“If the page needs a scroll before it says what it sells, the scroll is already optional.”
Speed is a conversion lever, not an engineering vanity metric
Every client site we take on gets audited before a pixel moves. Speed grades routinely sit at C or worse, almost always from unoptimised media and render-blocking third-party scripts. Fixing those two categories moves the grade to A without touching the design — and the revenue follows, because fewer people abandon before the page paints.
95+
Lighthouse score we ship as standard
3.4x
Average revenue lift across rebuilt funnels
C → A
Typical speed-grade move before any redesign
What not to do
- Rebuilding the visual layer while keeping the information architecture that confused people.
- Running an A/B test on a page that gets too little traffic to ever reach significance.
- Adding a carousel to avoid deciding which message matters most.
- Shipping a redesign without baseline numbers, so no one can prove what changed.
Final word
A CRO-led redesign is not a different visual style — it’s a different order of operations. Diagnose before you design, ship the structural fixes first, then make it beautiful with the constraints the data gave you. That sequence is why the sites we rebuild tend to hold their gains long after launch.
Atif Ali · WebPex Solutions
Writes about the parts of web design that show up in revenue reports.