CRO Program

On a mature product, a redesign by taste is a bet with real money on the table.

The client — a desktop-software maker with millions of visits a year — sells directly from its landing pages. At that scale even one percent of conversion is real revenue, so nothing ships on opinion: every change to a page goes out as an A/B test against the current version, and only a win at 95%+ significance stays.

Pattern, hypothesis, split, verdict — then again.

Each test starts from a pattern library built on hundreds of past experiments: what tends to win on pricing pages, on heroes, on forms. A hypothesis becomes a variant, traffic splits 50/50, and the numbers decide. Losers aren’t waste — they’re the map of what this audience ignores.

Four of the thirty — control on the left, winner on the right.

A second CTA in the hero — “Free Download” next to “Buy Now” — lifted conversions by 56% on 160,000 visits: from 10,825 to 16,782 orders. A rebuilt pricing block took the page from 26.7% to 64.9% conversion, a +143% lift. A full redesign of the product page added 81%. And a headline that leads with a number — “Your editing workflow just got 40% faster” — beat the descriptive one by 58%.

A year and a half of testing — and a page that earns measurably more than the one it replaced.

• 30 winning tests at 95%+ statistical significance
• +56% conversions from a single hero test —
  10,825 → 16,782 orders on 160,000 visits
• +143% on the pricing block, +81% on a full page redesign,
  +58% from a quantified headline
• Every losing variant documented —
  the pattern library got smarter with each round