What is cart abandonment?

Adding a product to the cart expresses a concrete intent to buy, firmer than a visit to the product page. If the session ends without a transaction, that intent stays on record and unpaid. The gap between the two is what the metric describes.

The analysis goes wrong at almost the same spot every time: confusion with checkout abandonment. These are two different stretches of the same journey. Cart abandonment covers everything that happens once the product is in the cart, including comparing prices in another shop or keeping the cart for another day. Checkout abandonment starts later, once the user has begun the payment process, and only covers the screens for details, shipping and payment.

In practice this means the two figures never match and are not interchangeable. Cart abandonment always comes out higher, because it also picks up people who use the cart as a wish list or as a calculator for the final price with shipping. Checkout abandonment is lower and more expensive: by then the user had already decided to buy and something stopped them. Mixing the two figures assigns the payment provider a problem that sat on the product page, or the other way round.

The metric is the counterpart of the completion rate. You count the units that reach each measurement point and divide them.

Abandono del carrito (%) =
(1 - purchase / add_to_cart) x 100

Abandono del checkout (%) =
(1 - purchase / begin_checkout) x 100

In Google Analytics 4 those points are ecommerce events with fixed names. add_to_cart marks the product entering the cart. begin_checkout marks the first step of the payment process. purchase marks the confirmed transaction. Between begin_checkout and purchase sit two intermediate events, add_shipping_info and add_payment_info, which show on which payment screen people drop out.

The choice that changes the result most is the counting unit. You can count events, sessions or users, and each option returns a different number. Counting events inflates the denominator, because the same person adds several items in one visit. Counting sessions breaks the cases where someone adds today and buys tomorrow. Counting users captures those delayed purchases, but requires a declared time window. No unit is the right one in the abstract; what does not work is switching it between two reports and comparing the results as if they measured the same thing.

The period also belongs settled before the calculation. A cart created on the 30th can convert on the 2nd of the following month, and that offset pushes the figure upwards in strict monthly closings.

The figure on its own decides nothing. What decides is where the journey breaks, and seeing that requires separating the two stretches.

If the large gap sits between the cart and the start of payment, the problem is usually in the information. An unclear final price, shipping costs that appear late, delivery times invisible before moving on. The sensible answer is work on the product page and the cart page, not on the payment provider.

If the large gap sits inside the checkout, the answer is the opposite. There you review the form, the obligation to create an account, the payment methods available and the validation errors that throw the user out. Polishing the product page in that scenario moves nothing.

The metric also sets the recovery budget. An abandoned cart is a user identifier with a specific product attached to it, and that is worth more than a cold advertising impression. Knowing how many carts are lost per month and what they are worth on average makes it possible to calculate whether a recovery sequence or a remarketing campaign pays off, instead of launching it out of habit.

• Measure the cart stretch and the checkout stretch separately, and report both figures together so nobody confuses them.
• Write the ecommerce event names exactly as the GA4 documentation defines them. A custom variant is recorded as a custom event and feeds nothing in the ecommerce reports.
• Show the total cost with shipping and taxes before asking for the first personal detail, so the price no longer changes inside the checkout.
• Allow purchase without creating an account and offer registration after the order is confirmed.
• Implement add_shipping_info and add_payment_info to know which specific payment screen loses users, instead of treating the checkout as a black box.
• Put the counting unit and the time window in writing, and repeat the same calculation before comparing periods.

• Taking a market average as your own target. Such averages aggregate sectors with very different decision cycles and say nothing about one specific shop.
• Calling cart abandonment what is in fact checkout abandonment, and then looking for the cause in the wrong part of the site.
• Comparing a month calculated by sessions with one calculated by users and presenting the difference as an improvement.
• Reading every increase as a failure. A campaign bringing cold traffic raises the figure without the checkout having got worse.
• Ignoring the effect of consent on measurement. Events that are not sent are missing on both sides of the division, and rarely in the same proportion.