What wins isn't campaigns; it's microcopy and the delays in the checkout step. Most of the work that reduces cart abandonment doesn't show up in a screenshot.
A discount is the most expensive fix
When cart abandonment shows up, the first reflex is a discount. But a discount pays for unresolved uncertainty out of your margin. You need to measure the uncertainty first: where users stop, what they're looking for, which field they fill in twice.
When we watched session recordings on Lipton's Amazon storefront, we saw that half the abandonments happened at the step where the delivery time wasn't shown. It had nothing to do with price.
Most orders lost in the cart are lost not to price but to uncertainty.
Five quiet decisions
None of the changes we made were campaigns:
The delivery time moved to the top of the cart. The shipping cost was shown upfront on the product page. The address form was cut down to a single screen, with autocomplete added. The checkout button changes its label while loading. Error messages appear as a single line under the field instead of a red banner.
Set up a measurement routine
To be able to defend changes like these, a measurement routine is essential:
- A step-by-step funnel. Every step from cart to confirmation is measured separately; the overall abandonment rate on its own tells you nothing.
- Field-level error logging. We log how many times each form field throws an error; the three fields with the most errors are reviewed every month.
- Real-device timings. The checkout step is measured with field data, not in a lab; we track the 75th percentile, not the median.
What we measured
The total impact over a ten-week period:
What we learned
The cart belongs to product, not marketing. When abandonment drops, your margin stays intact; conversion bought with a discount has to be bought again every month.



