Insights / Guide / Conversion rate optimisation (CRO)
GuideCheckout fixes that recover abandoned orders
Why online carts are abandoned, the checkout changes worth testing first, and how to run tests that measure orders rather than clicks.
Most online shops spend heavily to bring visitors to their site, then lose the majority of them at the final step. A shopper who has chosen a product, added it to the cart and started to pay has shown clear intent. Losing them at that point is the most expensive leak in the funnel.
Baymard Institute’s analysis of dozens of studies puts average online cart abandonment at 70.22%. Its benchmark of leading ecommerce sites also finds that the average large site could raise conversion by around 35% through better checkout design, and that about 65% of sites deliver a mediocre or worse checkout experience. In other words, the problem is common, and much of it is fixable.
This guide sets out where customers drop out, the fixes worth testing first, and how to test them so the results can be trusted.
Where customers drop out
Abandonment has many causes, and some are not worth fighting. Plenty of people add items to a cart to compare prices, save a list for later or check delivery costs with no intention of buying today. No checkout redesign will convert them.
The fixable causes cluster around friction and surprise. They tend to appear in roughly the same places on most sites:
- Extra costs revealed at the last step, such as delivery charges, taxes or fees that were not visible on the product page.
- Forced account creation before the customer can pay.
- Long forms that ask for information the shop does not need.
- Limited payment options, especially when a customer’s preferred local method is missing.
- Unclear error messages that do not say what is wrong or where.
- Uncertain delivery: options are listed but the actual arrival date is not.
- Slow pages or layouts that are awkward on a phone.
Your own analytics will show which steps lose the most people. Look at the drop-off between cart, delivery details, payment and confirmation, and split the figures by device. A step that performs well on a laptop and badly on a phone points to a mobile design problem rather than a pricing one.
Practical fixes to test
The following changes address the most common causes. Not all will help every shop, which is why each should be tested rather than simply rolled out.
Show the full cost early
Show delivery costs and any taxes or fees as early as possible, ideally on the product page or in the cart. A delivery cost calculator or a clear threshold for free delivery removes the moment of surprise. Customers are far more tolerant of a known cost than an unexpected one.
Offer guest checkout
Let customers pay without creating an account. Offer account creation after the order is confirmed, when the only extra step is choosing a password. Most customers who would have created an account will still do so, and those who would not are no longer lost.
Shorten the form
Remove every field you do not truly need. A second address line, a company name or a separate billing address can be optional or hidden until requested. Use address lookup from a postcode, and set the right keyboard on mobile for numbers and email addresses. Each field removed is one less reason to give up.
Offer the right payment methods
Offer the payment methods your customers actually use in each market: cards, digital wallets, bank transfers or instalment options, depending on the country and the basket size. Check your analytics for the countries and devices your customers use, and match payment options to them.
Make errors specific and fixable
When something is wrong, say exactly what and where, next to the field, in plain language. Keep everything the customer has already entered. Clearing a form because one field was invalid is one of the fastest ways to lose an order.
Promise a date, not just a method
Show the expected delivery date rather than only “standard” or “express”. A customer deciding whether a gift will arrive in time needs a date, not a description.
How to test properly
Checkout tests are easy to run badly. A change that looks positive after three days can turn out to be noise once a full week of traffic, including the weekend, has passed. Good practice is simple but requires patience.
- Change one thing at a time, so you know what caused the result.
- Estimate in advance how much traffic you need to detect a meaningful difference, and run the test until you reach it.
- Run tests over whole weeks to capture weekday and weekend behaviour.
- Measure completed orders and revenue per visitor, not only clicks or step completions.
- Record every test, its result and what you learned, including the ones that failed.
Keep a backlog of ideas ranked by expected impact and effort. Start with changes that are cheap and address the biggest drop-off in your own data. A team that runs two well-designed tests a month learns more in a year than one that runs a dozen rushed ones.
Smaller shops and testing
If your traffic is too low for reliable A/B tests, make one clear change at a time, compare full months before and after, and watch for seasonal effects. Usability sessions with five or six real customers can also reveal the worst problems quickly.
Recovering carts that are already abandoned
Even a well-designed checkout will lose some customers. For those who have given an email address and agreed to hear from you, a short reminder can bring a share of them back. Keep it simple: one message within a few hours showing the items left in the cart, with a direct link back to the checkout and the cart still intact.
Be careful with discounts in these messages. Offering money off to everyone who abandons a cart teaches regular customers to abandon on purpose and wait for the code. If you test an incentive, limit it to first-time customers or higher-value baskets and compare the margin, not just the recovered orders. A clear reminder about delivery dates or returns often works as well as a discount, at no cost.
Above all, respect consent and frequency. One or two reminders are helpful; a week of them damages the relationship and your sender reputation.
Beyond the checkout
The checkout is where intent turns into revenue, but it is not the only place customers are lost. Site search that returns poor results, product pages without clear sizing or stock information, and vague delivery promises all affect whether a visitor reaches the checkout at all.
Treat the whole journey as one funnel. Map the drop-off at each stage, from landing page to confirmation, and fix the biggest leak first. Often that is the checkout. Sometimes it is search or product pages, and fixing those sends more qualified shoppers to a checkout that is already working.
Where to begin this month
Complete a purchase on your own site using a phone, as a new customer, with a delivery address outside your home city. Note every moment of hesitation, surprise or extra effort. Then check your analytics for the step with the highest drop-off. Choose one fix from the list above that addresses it, test it properly over at least two full weeks and measure orders. That single test will tell you more than a full redesign proposal.
Source: Baymard Institute, cart abandonment rate research and checkout usability benchmark, 2025.
