Why people abandon the basket and checkout, and what to test

The basket is the busiest graveyard on most ecommerce sites. Baymard Institute averages 50 published studies to put the documented cart abandonment rate at around 70%, and that number turns up in a lot of agency pitches as proof that there’s money lying on the floor. Some of it is. A good share of it isn’t, and knowing the difference is most of the job when you’re deciding what to test.

Reading the abandonment number properly

Plenty of people put things in a basket with no intention of buying that day. They’re saving items for later, comparing prices, or checking what delivery will cost. In Baymard’s survey of US online shoppers, 42% said they had abandoned a basket because they were just browsing or weren’t ready to buy. No checkout redesign will change that, and an agency that promises to recover most of the 70% is selling you a number nobody can deliver.

When Baymard sets the browsers aside and asks people who meant to buy why they left, the reasons are more useful. 40% said extra costs such as delivery, tax or fees were too high. 20% said delivery was too slow. 19% didn’t trust the site with their card details. 18% left because the site wanted them to create an account. 17% found the checkout too long or complicated, and another 17% hit errors or crashes. 13% weren’t happy with the returns policy, 12% couldn’t see the total cost up front, 10% had a card declined and 9% didn’t find enough payment methods.

Treat those as what they are, a survey of what people say about past behaviour. They’re a good map of where to look, and a poor estimate of what any single fix will be worth on your site. Your own evidence decides the order.

What usually goes wrong

Surprise costs top the list for a reason, and the problem usually starts upstream. Baymard’s product page research found 67% of sites don’t show an estimated total, delivery included, near the buy button. If the first time someone sees delivery is the basket, the basket gets blamed for a product page decision.

Forced account creation is the second big one. Baymard reports that 24% of US shoppers abandoned an order in the previous quarter solely because a site made them create an account, and that 47% of sites that do offer guest checkout don’t make it prominent enough. A guest option shown as a small text link below a sign-in form is easy to miss, and a missed option might as well not exist. Baymard’s guidance is a clearly labelled guest button at the top of the step, using the word “guest”. You can still offer an account after the order, when the customer has a reason to want one.

Delivery and payment are the rest. Slow or unclear delivery options, a single delivery date with no choice, missing payment methods the audience expects, or a card form that throws errors on a phone. The field count matters here too, and the lesson on forms covers it in detail.

Where to look for evidence

Build the funnel step by step, basket to delivery details to payment to confirmation, and break it down by device, by new and returning customers, and by payment method. A big drop at one step on one device is a much better test lead than the overall abandonment rate. Look at error logs and declined payments, which are often owned by a different team and rarely make it into CRO conversations. Ask customer service what people ring up about mid-order. A short exit survey on the basket page, asking what stopped people today, is cheap and can tell you whether the Baymard reasons apply to your customers or only to someone else’s.

What is worth testing

Show the delivery cost, or the free delivery threshold, before the basket. Make guest checkout the first, most obvious choice. Show the full total, including delivery, on the basket page and keep it visible through checkout. Offer the payment methods your customers already use, which you can check from your payment provider’s data before you test anything. Test a clearer delivery choice, with dates rather than vague day ranges.

Checkout tests need careful reading. Conversion is the obvious primary metric, but watch average order value, payment failures and, where you can, returns and repeat purchase. Some changes shift timing more than behaviour, so a test that moves a few people from tomorrow to today can look like a win and fade once it ships.

The versions to refuse

Checkout is where the most damaging tricks live, because people are close to paying and less likely to notice. Pre-ticked boxes that add insurance, gift wrap, donations or a subscription. Fees that appear at the last step, labelled “service” or “handling”, that weren’t in the price shown earlier. A guest option hidden on purpose because the business wants the email address.

These will often win a short test on revenue per visitor, which is why they need refusing in principle. In the UK, regulation 40 of the Consumer Contracts Regulations 2013 says a trader can’t charge for extras without the customer’s express consent, and that a pre-ticked box doesn’t count as consent. The Competition and Markets Authority’s price transparency guidance, published in November 2025, covers drip pricing and mandatory fees, which belong in the price people see first. The cost of getting it wrong turns up in chargebacks, complaints, regulator attention and customers who don’t come back.

What to ask whoever builds the test

Ask which abandonment reasons apply to your own customers, and what evidence shows it. Ask whether the funnel is measured step by step and by device. Ask how payment failures and errors are tracked, and who owns them. Ask what the guardrails are, beyond conversion. If a proposal includes anything ticked by default, or a fee that isn’t in the first price shown, ask for it to be removed before you discuss the rest. And if the pitch leans on recovering the 70%, ask how much of that number they think is people who were only browsing.

Put it to work

Go through your own checkout as a new customer, on a phone, with one product in the basket. Write down the total you expected from the product page and the total you were asked to pay, every point where you were asked to create an account, and every extra that was added or offered without you choosing it.

Your answer

Show the worked example

There is no right answer. Write yours first, then compare.

Sources

1. Baymard Institute: Cart Abandonment Rate Statistics
2. Baymard Institute: Make Guest Checkout the Most Prominent Option
3. Baymard Institute: Product Page UX, 10 Pitfalls and Best Practices
4. Competition and Markets Authority: Price transparency guidance (CMA209)
5. The Consumer Contracts (Information, Cancellation and Additional Charges) Regulations 2013, regulation 40