Everything else in a shop is persuasion. The checkout is not. By the time somebody arrives there they have decided, and your only job is to get out of the way.
That makes it the one part of a shop where you can improve results without being cleverer about anything. You are not writing better copy or taking better photographs. You are removing obstacles from in front of people who already want to pay you.
Most checkouts were configured once, by whoever built the shop, using whatever the platform offered by default, and never looked at again.
Forcing an account is the classic one
Requiring somebody to create an account before they can pay is the single most reliable way to lose a first-time customer, and it is still switched on in a great many shops.
The reasoning is understandable. You want the account so they can track the order, reorder easily, and be marketed to later. The trouble is that you are asking for a commitment before you have earned one, from somebody who has never bought from you and does not yet know whether they will again.
Turn on guest checkout. Then, on the confirmation page and in the confirmation email, offer to save the details they have just typed as an account with one click. You get most of the accounts you would have got anyway, from people who have already had a good experience, and you stop losing the ones who would simply have left.
Every field, justified out loud
Go through your checkout and, for each field, answer one question: do you need this to take the money and get the goods to the right place?
Some things fail that test immediately. Company name, when most customers are individuals. A second address line, when the first is usually enough. Title, which almost never matters and occasionally annoys. A landline and a mobile. Date of birth, unless you sell something age-restricted. “How did you hear about us”, which belongs in a survey after the sale, not between somebody and their card.
Others can be inferred rather than asked. County is usually derivable from the postcode. So, in most cases, is the whole address, and a postcode lookup turns five fields into one and a dropdown.
The ones that survive are the ones you genuinely cannot fulfil the order without. That is normally name, email, delivery address, and payment details. It is a shorter list than most checkouts contain.
Count the taps, not the fields. On a phone, a dropdown with thirty options is far more work than a text field, and a date picker is worse than both. Complete your own checkout on your own phone, one-handed, and count every tap and every keyboard change. That number is what your customers are actually paying to buy from you.
Payment: offer the ones people already trust
Card details are the slowest thing anybody types on a phone, which is why the wallets matter so much. Apple Pay and Google Pay turn the longest step in your checkout into a fingerprint, and for a lot of shops they now account for a substantial share of mobile orders.
PayPal remains worth offering, not because it is elegant, but because a certain kind of customer trusts it more than they trust an unfamiliar shop with their card. Removing it to tidy up the page loses those people quietly.
Buy now, pay later is a genuine decision rather than an obvious yes. It lifts orders for higher-value goods and it suits some audiences much better than others. It also comes with fees and with a duty to think about whether encouraging credit fits how you want to treat your customers. If you do offer it, present it as one option among several rather than the headline.
Be honest about costs early
Nothing empties a basket faster than a total that grows at the last step. Postage that only appears once an address has been entered, or a surcharge that turns up at payment, feels like a trick even when it is not intended as one.
Show the delivery cost, or the rule that determines it, before somebody starts the checkout. What the charge should actually be is a subject of its own later in this series. Whether it arrives as a surprise is a checkout question, and the answer is that it never should.
Errors that do not punish people
Something will be typed wrong. What happens next decides whether the order survives.
Check each field as they leave it rather than only when they press the button, so the message arrives next to the problem while they are still thinking about it. Say what is actually wrong in plain words, not “invalid input”. Never clear what they have already typed, which is the behaviour most likely to make somebody give up entirely. And if the card is declined, keep everything else and let them try another card without starting again.
Small things that add up
Let the phone help. A field expecting a number should bring up the number keypad, a field expecting an email should bring up the keyboard with the at sign, and fields should be labelled so the browser can autofill an address it already knows.
Show progress, so somebody on step two of three knows there is an end in sight.
Keep the reassurances where the doubt is. A line about returns, a padlock near the card fields, and your phone number visible in case something goes wrong all cost nothing and are most valuable at exactly this moment.
Take away the exits. A checkout does not need your full navigation menu, a newsletter sign-up, or a panel of related products. There is one thing to do on this page.
Where to start
Buy something from your own shop. Not a test order through the admin panel, a real one, on your own phone, on mobile data, paying with a real card. Then do it again as a first-time customer with no saved details.
Write down every moment you hesitated, every field you resented, and every tap that felt unnecessary. That list, in order, is your work.
If you would like somebody to go through a checkout properly and tell you which of the defaults are costing you orders, get in touch. Setting up the practical machinery of selling, from payments to delivery options, is part of how we build and look after an online shop.