If the delivery date appears under the buy button, mobile visitors will add to cart more often, because delivery time is the second most cited reason for leaving in our exit survey.
Nothing here blocks the launch. An unverified check says what it would take to verify it.
A simplification or a cost saving isn’t trying to do better. It only has to do no worse. On “Add to cart” (8.2%), here’s what each margin costs, with a single look at the end, if the variant actually performs as well as the original. The test doesn’t declare a winner.
As a primary metric, a retention metric like “returns between day 7 and day 14” only counts a visitor 14 days after exposure. To detect 10%, the test goes from 20 to 34 days, with the same sample size.
Targeted pages /produits/*, devices: mobile. No excluded pages, entire audience, activated on page load.
Over the last 7 days, 33% of measured new visitors (from 32% to 34%, 2,030 of 6,155) landed on a page targeted by this test. Those who accept cookies while browsing see the original on that first page: their variant and exposure start on the next targeted page, and those who leave before are not counted.
Assignment does not depend on when consent is given, so the variants stay comparable. Returning visitors see their variant as soon as the page loads.