The checkout that stops losing carts: a 10-point audit

Customers abandon carts for reasons that can mostly be named and fixed: costs shown too late, a forced account, a form that is too long, a payment page that does not feel trustworthy, and a missing payment method the customer expected. Baymard Institute documents an average abandonment rate of 70.22% (a meta-analysis of 50 studies, as of 22 September 2025), but that number misleads if taken literally: a chunk of abandonment is natural “just browsing”, unrecoverable by any fix. What is genuinely recoverable is the part driven by friction in the process. So we treat the checkout as an engineering and process problem, not a marketing trick. Below is a 10-point audit, each with what to check, the typical mistake and how to fix it.

First separate fixable abandonment from natural abandonment

Before you fix anything, split abandonment into two piles. The first is shopping behaviour you do not control: 42% of shoppers in Baymard’s research abandoned because they were “just browsing, not ready to buy”. Comparing prices, using the cart as a wish list, checking the delivery cost is normal traffic you will not remove. The second pile is abandonment for concrete, repeatable reasons, and that is where the game is played. Among those who abandoned for a reason other than browsing, Baymard reports the most common reasons (respondents could select several, so the shares do not sum to 100), and that list is practically a ready-made audit plan:

• 40%: extra costs too high (shipping, tax, fees).
• 20%: delivery too slow.
• 19%: did not trust the site with card details.
• 18%: forced account creation.
• 17%: checkout too long or complicated.
• 17%: site errors and crashes.
• 13%: unsatisfactory returns policy.
• 12%: could not calculate the total cost upfront.
• 9%: not enough payment methods.

The order of the points below is not random: it runs roughly from the cheapest, most common causes to the more technical ones. Walk through it as a checklist for your own store.

1. Number of steps and form fields

What to check: count the screens and fields from cart to confirmation, and separately the fields actually required. Every extra field and every extra screen is a point where someone drops out.

Typical mistake: fields nobody needs to fulfil the order (a second phone number, a tax ID required from a private customer, a mandatory “company” field), plus a form that blocks the browser’s autofill. This is exactly the 17% “too long or complicated checkout” in Baymard’s data.

How to fix: remove every field you will not genuinely use in fulfilment, put the rest in a single column, enable autofill and show a progress bar so the customer sees how much is left. An address is one field with suggestions, not eight separate boxes.

2. Guest checkout versus a forced account

What to check: whether people can buy without creating an account, and whether guest checkout is the default rather than hidden.

Typical mistake: a registration screen set as a gate before payment, with a password, terms and email confirmation before the customer has even paid. This is a distinct, well-documented cause: 18% abandon because the store forced them to create an account.

How to fix: make guest checkout the default path and offer the account after the purchase, in one click, using the data the customer already provided. An account should be a reward for buying, not a tax before it.

3. Costs visible early, no surprises

What to check: at what point the customer sees the full cost including delivery and any fees, and whether the last step surprises them. This is the single most common cause in the whole study: 40% abandon over extra costs being too high, and a separate 12% because they could not calculate the total order cost upfront.

Typical mistake: delivery cost and fees revealed only at the last step, right before payment, once the customer already feels they “almost bought”.

How to fix: show the delivery cost in the cart (a calculator or a free-shipping threshold with a clear “you are X away from free delivery”), do not add fees at the end and do not hide anything behind an asterisk. If you must add a fee, name it and show it immediately.

4. Payment methods matched to your market

What to check: whether you offer the methods your customers actually expect, and whether they are within reach rather than buried under “other”. Baymard records 9% of abandonment from too few payment options. Market reality in Poland: BLIK is now one of the core online payment methods, alongside pay-by-link fast transfers and Apple Pay and Google Pay wallets, and on higher-value carts buy-now-pay-later and instalments are growing.

Typical mistake: a checkout offering only a card and a traditional transfer, or showing BLIK only after expanding a list.

How to fix: surface BLIK and fast payments as visible, default options, add mobile wallets, and on pricier ranges consider instalments or deferral. Do not guess the split: the data on how your customers pay lives in your own payment gateway.

5. Validation and form error handling

What to check: how the form behaves when the customer makes a mistake. When it flags the problem (inline or only after submit), whether the message says specifically what to correct, and whether data survives the error. This is part of the 17% “site errors and crashes”, and the most frustrating kind of friction, because the customer wanted to buy and got punished.

Typical mistake: validation only after clicking “pay”, half the fields wiped, a red border with no explanation, and an intolerant format (a phone number rejected for a space, a postcode for a missing hyphen).

How to fix: validate inline on each field, write the message next to the field it concerns, never clear entered data, and accept varied formats for phone, tax ID or postcode instead of forcing the customer to guess yours.

6. Performance and handling on a phone

What to check: how fast the checkout loads on a phone on a weaker connection, and whether it can be operated comfortably with one thumb. Since most store traffic is now mobile, a checkout designed “for desktop” loses customers where they actually buy.

Typical mistake: a heavy, slow final step, a keyboard full of letters where a code or number goes, buttons too small for a finger, and fields that hide behind the keyboard.

How to fix: strip the checkout to the minimum, set the right field types (a numeric keyboard for codes and phone numbers), make touch targets large enough, and test the whole flow on a real, inexpensive phone, not just in a preview on a big screen.

7. Trust and security at the payment step

What to check: whether the payment page looks secure and consistent with the rest of the store. 19% of shoppers abandon because they do not trust the site enough to enter card details, and a separate 13% bounce off an unclear returns policy.

Typical mistake: an abrupt jump to a foreign-looking gateway with no warning, no security signals at all, typos and half-finished translations, no contact details and a hidden returns policy.

How to fix: keep a consistent look across the whole path, use recognisable payment gateways, show a short note about payment security, expose visible contact details and a link to the returns policy next to the pay button. Trust at this step is built from small things and lost by one unsettling signal.

8. Address and delivery: methods the customer knows

What to check: which delivery methods you offer, how fast you deliver and how easy it is to enter or pick an address. Delivery being too slow is the second most common cause in the study (20%).

Typical mistake: no parcel locker in a market where it is one of the core pickup options, laborious manual entry of a long address with no suggestions, and no clear date or cost before choosing.

How to fix: add pickup-point or locker selection with a search or map, enable address autocomplete, and for each method show the cost and a real delivery date immediately. A customer decides more calmly when they see “tomorrow to a locker for X” than when the date only appears after typing everything.

9. Stock and reservation during purchase

What to check: whether the stock shown in the checkout is current and whether the item is reserved for the duration of checkout, plus what happens if it runs out mid-flow. This is the purely engineering part of the audit and the most painful kind of abandonment, because it happens after the decision to buy.

Typical mistake: selling the last unit to two people at once, no cart reservation, and a hard error on the payment screen instead of information.

How to fix: sync stock with a safety buffer, reserve items for the duration of checkout, and instead of a hard error show “only a few left” early. We describe the mechanics of safe stock syncing separately in the piece on B2B store and ERP integration; the same principles protect the checkout in any store, not just a wholesale one.

10. Measuring the funnel step by step

What to check: whether you even know which step loses your customers, or only one overall abandonment rate. This is the point worth starting from in practice, even though it is last on the list.

Typical mistake: a company knows its “cart abandonment percentage” but not whether the drop happens on details, delivery or payment, so it fixes blind and everything at once.

How to fix: set events on every checkout step (entering the cart, details, delivery, payment, confirmation), see where the curve falls hardest, and only then point your fixes there. Change one thing at a time and measure the effect, otherwise you will never learn what worked. Measurement turns the ten-point list from generic advice into your own, prioritised set of tasks.

Where to start, and when it is not a fix at all

A practical order: start with point ten (measurement) to know where you actually lose people, then the cheapest and most common wins, meaning costs shown early (3), buying without an account (2) and the payment methods your market expects (4). The rest is polishing form, trust and operations. There is a limit worth stating plainly, though. If at successive points you keep hitting “we cannot change that in our platform” (you cannot shorten the steps, force guest checkout or add a payment method), this is no longer a checkout fix. It is one of the signs a store has outgrown its platform, which we cover separately, and a rigid checkout is often the most expensive of them, because the lost sales are hard to even see in reports. In our case that diagnosis, together with the arithmetic of whether the fix fits the current platform, happens during the Blueprint.

A checkout that stops losing carts does not come from one trick or an exit-intent discount popup. It comes from ten places where the process either respects the customer’s time and trust or chips away at them. The good news is that all ten can be checked in a single afternoon with this list in hand, and most fixes are not a revolution, just the removal of friction someone once added without meaning any harm.