Cart abandonment statistics

Calculate one defined cart rate

Cart abandonment rate = abandoned eligible carts / eligible carts created × 100. Count distinct carts, rather than item additions. For your own report, define an abandoned cart as an eligible cart without a linked qualifying completed order by the end of its follow-up window.

Choose a cart identifier, a cart-creation period and a fixed time allowed for completion after creation. State when a changed or restored cart becomes a new cart. Compare cohorts only after each has completed that follow-up. Apply the same bot and test exclusions throughout; if there are no eligible carts, leave the rate unavailable.

Improve the product information shoppers return to: audit the product gallery.

[8]

Keep cart and checkout denominators separate

Shopify distinguishes adding an item from starting checkout. Use the working definitions below to keep those stages separate, and record how each cart or checkout links to an order. A reminder flow has another denominator: people eligible to receive that message.

Choose the completion event consistently, including how cancelled or refunded orders affect it. These operating definitions need not match the native report in your platform, so reconcile its event rules before comparing rates.

Keep recovery creative aligned: check Shopify variant images.

[8]

Fix the failed step before increasing reminders

Inspect total cost, delivery timing, payment availability and guest checkout before expanding recovery messages. Check that the cart shows the selected variant and makes bundles clear. A browsing shopper and a shopper blocked by a reproducible payment error need different interventions.

Use the reason evidence to prioritize investigation. Its percentages cannot be subtracted from the pooled abandonment rate because their denominators differ. Test the affected step against a comparable eligible cohort and report qualifying completed orders with the same cancellation and refund rules.

Use the AI product photography guidance to resolve visual uncertainty before testing recovery messages.

[7][1]

Investigate the friction your funnel actually records

Baymard’s extra-cost finding points to a cost-visibility test, not an image test. Compare the current experience with a clear total-cost estimate near the product’s buying controls for destinations where costs can be calculated. Preserve the product, offer and payment choices.

Link distinct cart IDs to checkout starts and qualifying orders. Keep initial completion and later retained orders as separate outcomes, so refunds do not silently change the purchase event. A gallery correction belongs in a separate test when the selected color, bundle contents or product detail is the observed problem.

[1][8]