<b>The CRO test that 'failed' because of returning users</b>

<b>The CRO test that 'failed' because of returning users</b>

A subscription box ran a bold new homepage and called the test a loss: conversion dropped 0.4% overall, so they reverted within a week.

The clue surfaced when someone split the test by new versus returning visitors. New users converted 1.2% higher on the new design. Returning users, trained on the old layout, fumbled and dragged the blended number negative.

They re-ran it serving the new design only to first-time visitors, where the redesign was actually meant to win.

New-visitor signup rate rose from 3.1% to 4.3%, and the 'failed' design shipped to exactly the audience it helped.

The lesson: returning users carry muscle memory that poisons redesign tests. Segment new from old, or a winner dies under the weight of habit.