Testing a First-Order Discount Message on Cards

New visitors may be finding the first-purchase offer too late if it only shows up in the cart, if new customers say in surveys that they did not know about it, or if new visitors open products from the list and leave without adding anything. You would expect those who see the offer on the card to add more to the cart and place a first order more readily. If the offer itself is weak, work on that first; before launch, make sure the card shows the price and terms accurately and states that the offer is for new customers only.

Post-discount revenue per eligible new visitor

Divide post-discount revenue by all eligible assigned new visitors. If the reported revenue is already net of discount, do not subtract the discount again.

Also track

• New-customer purchase conversion rateCheck whether first purchases increase.
• Add-to-cart rateCheck whether more visitors who see the offer add an item.
• Gross marginMake sure discount cost does not erase the gain.
• Second-order rateCompare second orders over the same follow-up period among assigned users.

• Early visibilityShow the existing first-purchase offer on the product card.
• EligibilityDefine eligible new users before assignment and apply the same rule in both arms.
• Keep constantKeep discount amount, terms, prices and redemption flow unchanged.
• DeviceReview mobile and desktop results.
• Separate testsTest copy placement and percentage-versus-currency presentation separately.

1. 1Even if new-customer purchase conversion or the add-to-cart rate goes up, post-discount revenue per eligible new visitor is what decides the test. If gross margin erodes, more first orders do not make the change worth shipping.
2. 2Keep the result to this category list and this offer, and look at mobile and desktop on their own. Report the second-order rate for both arms over a follow-up window of equal length, and keep in mind that unchanged repeat buying does not cancel out a profitable first-order contribution.
3. 3Order values vary from one customer to the next, so a small gap in revenue per visitor can easily stay unclear. Before launch, enter the current new-customer conversion rate and the smallest lift you care about in the sample size calculator. Divide the users it asks for per variant by daily eligible traffic per arm to see the run length; a revenue gap usually settles later than a conversion gap, so do not report an unclear one as a win.
4. 4To tie any difference to the card message, leave the campaign area and the discount itself unchanged for the whole test. Throughout the test, confirm with mobile and desktop screenshots that the message keeps the card's visual hierarchy intact and does not overlap the product image; a difference measured on a broken card may come from the layout, not the message.