Checkout page conversion rate: orders per start

The rate divides two counts that often come from different places. The checkout start may be recorded when the address page loads, which includes a shopper who clicked by mistake. The order may be recorded on a confirmation page that some shoppers never reach, such as one who closed the tab after paying. A shopper counted at the start who never orders is a lost sale. A shopper who paid but never saw the thank-you page is a counting gap, and it moves the rate just as much. Compare both counts against the same orders in your store admin before you read the rate.

• Checkout starts and placed orders for the same dates, pulled from the store admin
• The address form: does it accept a valid address from your main ad country without an error?
• Shipping and tax: do they appear before the payment step, and does the total stay the same?
• Payment options: is the method most of your shoppers expect shown on a phone?
• The step after payment: does the confirmation page load and record the order?

Before the checkout rate means anything, the two numbers need one origin. Take the orders from the store admin, and take the checkout starts from the same system for the same date range. If the checkout start is counted in an analytics tool, check how often it fires for one shopper who moves between the cart and the address page. Two counts from two tools will never match exactly, so decide which one is the denominator, write it down, and compare the same pair each month. A rate that changes because the counting changed is not a result.

Most checkout losses sit in a short list of places. The address form may reject a postal code format from another country, or require a line that is not needed. The total may change after the address is typed, because tax or shipping appears late. The payment page may show no option the shopper uses, or show one in a way that looks unfamiliar on a phone. Each one looks like the same drop in the rate. Walk the checkout yourself on a phone, in a private window, with a real address from your largest ad market, and write down the first moment you hesitate.

Checkout flow testing

A simulated shopper can reach the payment step and stop there, which covers the address, the delivery choice and the total. It cannot pay with a real card, sign in with Google or complete a payment that a bank has to approve. Test the payment step yourself with your payment provider's test mode, and check the confirmation page. A shopper who paid and then saw an error is a lost order, and the rate records it as a loss.

Shipping rules, tax settings and payment options change without a page redesign, and any one of them can add a field, move a total or hide a method. Run the checkout again after each change, from the same address, on a phone. When shoppers must sign in, check the checkout behind a login with a test account. Give each shipping country its own run, since a checkout that works for one country can stop at the address for another.