The checkout nobody checked after launch
We ran an audit on a site recently that, by most measures, is a good one. Fast-loading, well-structured, the result of real investment over several years. We found 56 empty links with no accessible name, 37 images with no alt text, and a heading hierarchy that put the main page title after twelve navigation headings. Focus indicators had been removed via a JavaScript workaround that a single line of modern CSS had made obsolete years earlier.
None of it was introduced deliberately. It was what happens when a site runs for a few years without anyone looking at it properly.
Checkout is usually where this matters most.
What checkout debt looks like
Checkout is the highest-traffic, highest-value sequence on the site. It’s also the page teams most consistently stop reviewing once it’s live. The logic is understandable: it works, it converts, there are bigger priorities. But conversion debt compounds quietly. A form field that lost its label in a template update. A CTA button whose contrast ratio fails on certain devices. A guest checkout flow that accumulated two extra steps somewhere between the original build and the version live today, and nobody noticed because the conversion rate held roughly flat.
The VAT toggle that screen readers can’t use. The payment error message that gives a code, not an explanation. The mobile keyboard that covers the field it’s supposed to be filling in. These aren’t edge cases. They’re the kind of thing that surfaces in the first ten minutes of a proper review.
In our experience, running optimisation experiments on top of foundations that have quietly drifted produces smaller returns than the test results suggest. The lift from a better-designed flow is real, but it competes with drag from things that were broken before the test started. Because the drag is invisible – spread across devices, browsers, and user types – it never makes it into the report.
What a proper look actually involves
It doesn’t need to be a project. Start with an hour.
Pull up the site on a device you don’t normally use. Go through the checkout as a first-time customer – from product page to order confirmation, without skipping ahead. Look for the point where you hesitate, the moment something isn’t obvious. Run an accessibility check on the checkout pages specifically: heading structure, form labels, focus behaviour, button contrast. Check whether the guest checkout option is still as prominent as it was at launch. Ask when anyone last tested the full journey on a slow connection, and what happens when a payment fails.
The list that comes out of that session will have things you can fix in a day and things that need more thought. Some of it will be mildly embarrassing. That’s the good version – findings caught internally, before a customer abandons at the point they were about to convert.
Why this falls through the gap
Heroics carry a cost beyond the obvious one – work that doesn’t demand attention loses to work that does. A checkout review doesn’t scream. It waits.
Going live is the start of building something, not the end of it. The review is what keeps it honest. If you’re in a service delivery relationship with your agency – measured on what ships, not what’s maintained – this kind of review won’t naturally find its way into scope. Worth knowing, and worth changing.
The checkout that’s been running since launch is the most likely place to find easy gains. Most teams just haven’t looked.