Cart abandonment is the share of shopping carts that are created but never turn into a paid order. The shopper picked items, started the purchase, and left before the money moved.
Working the rate out from two numbers
Divide paid orders by carts created, subtract from one, and express it as a percentage: 1,000 carts producing 300 orders is a 70% abandonment rate.
The denominator decides what the number means. Count every cart, including those opened by price-checking and by bots, and the rate looks catastrophic. Count only carts that reached the checkout and it looks flattering. Neither is wrong; using both interchangeably across a year is.
What is cart abandonment really counting?
A large share of what dashboards call abandonment was never a decision. The buyer pressed pay, the issuer refused, the page showed a generic error, and the session ended. That is a payment failure wearing the costume of a change of mind.
The two need separate treatment. A shopper who left at the shipping screen may respond to a reminder; one whose card was refused for insufficient funds needs decline recovery and retry logic, and will find an email about their forgotten cart insulting.
Why do customers abandon carts?
Cost, friction and doubt, roughly in that order. A shipping charge, tax or cross-border fee appearing only on the final screen changes the deal the buyer had already accepted in their head, and it is the most common single trigger.
Friction is the next layer: an account gate, a long address form, a missing local payment method. Doubt covers the rest — an unfamiliar seller name, no visible refund terms, no delivery date. Plenty of shoppers also use the cart as a wish list and never intended to buy today.
What a recovery email is really worth
Abandonment emails do recover orders, and their reported performance flatters them, because a share of those buyers would have come back unprompted. Judge the sequence against a holdout group that receives nothing, or it will take credit for sales it did not cause.
Send fewer than the tooling suggests, send the first within a few hours while intent survives, and never send one to a customer whose payment failed for a technical reason.
Chasing benchmarks is a poor use of time
Published industry averages blend desktop with mobile, one-page flows with multi-step ones, considered purchases with impulse buys. Your own rate last quarter, split by device and country, is the only comparison carrying information.
Fix it at the source instead: honest totals early, the methods the market uses, fewer fields. That is ordinary checkout optimization work, and it outperforms any email sent after the fact.