The Best Kind of Platform Upgrade is the One No One Talks About

A surprising amount of what your browser downloads when you click "buy" online exists for one reason: to keep an old version of Internet Explorer happy. Microsoft retired the browser in 2022. The code written to work around it is still there on countless sites, still loading, still costing. 

Internet Explorer is the most visible example, but old code piles up for plenty of other reasons. Frameworks lose their maintainers, libraries get superseded, and the assumptions a system was built on quietly stop being true. Often everything keeps working, so it gets left alone. What eventually pushes a business to act on it tends to come down to two things: security and performance.

That was the situation at a long term client of ours, a high volume e-commerce platform serving over a 1000 branded microsites. HappyPorch has worked with them since 2016, so when the time came to rewrite the checkout, we knew the platform, the business, and what was at stake.

The checkout itself was built seven years ago, in an era when developers spent a meaningful chunk of every project incorporating polyfills, small pieces of extra code that give old browsers the ability  to do things newer ones could already do, so that anything would function in IE at all. The framework underneath had since been deprecated, meaning its makers had stopped maintaining it and stopped issuing security updates, which is a real risk on a platform handling payments. 

The browser landscape had also changed completely. 

Three engines now power almost everything people use online, Chrome and Edge (which share the same underlying engine), Safari and Firefox. Crucially, they now all update themselves quietly in the background. That wasn't true a decade ago. Back then, plenty of people stayed on whatever browser came with their computer for years at a time, gradually falling further behind on features and, more importantly, on security. Today most users are on a recent version without ever thinking about it. And, rather than fight amongst themselves, new browser features are agreed across standards bodies. The audience had shifted too. Mobile was now 68% of users. Desktop had dropped to 30%.

Tom led the rewrite. He hadn't written the original code, and that matters. Often, the actual job, when you're modernising something business critical, is taking ownership of a codebase someone else built years ago, in different conditions, for different users, and making it yours to maintain confidently. That's most of what this work is.

Tom started by looking at the company's own analytics from the previous six months. He wanted to know who was actually using the checkout before he decided what to strip out. The answer was clear: almost everyone was on a modern, auto-updating browser. The legacy support wasn't useful to anyone. It was just sitting in the way.

What came out the other side is what you'd want from a clean rewrite. Blocks of code that ran to 30 lines in the old checkout are now often a single line. The production bundle, the JavaScript every browser downloads on every visit, dropped from around 950KB to 650KB. That's roughly a 30% reduction. Multiplied across more than a thousand client sites, and across the peaks of Black Friday and Christmas when these checkouts are running flat out, the cumulative data saving is real, even if the carbon maths is hard to pin down precisely.

What the end user notices is nothing. The checkout looks the same and behaves the same, the gains are structural. Our client now sits on a platform with improved security, no longer expensive to maintain, and no longer a brittle foundation under any future change they want to make.

There's also something to say about how it's going out. Rather than flipping every site over to the new checkout on a single day, HappyPorch can dial up the rollout site by site. A small cohort goes first, confidence builds, the next group joins. If something breaks, it breaks small. Our client’s customers don't need to do anything, don't need to be briefed in advance, don't need to migrate anything themselves. After the fact, the message is simple: we've made security and platform improvements. 

That kind of gradual rollout isn't always possible. The way the rewrite was scoped and architected is what made it possible, and it's worth recognising as deliberate engineering practice rather than as a happy accident.

The work itself isn't dramatic. There's no new feature to demo, no visible redesign, no announcement to make to end users. It's the kind of project that quietly removes risk, reduces cost, and clears the ground for whatever comes next, on a platform that's moving millions of British Pounds through it across a busy season. That's one type of work HappyPorch is comfortable doing: substantial, business-critical platforms with real legacy underneath them, where the value comes from doing the unglamorous thing carefully and at scale.

This is one strand of what we do. Our work with circular businesses runs from smaller, focused engagements through to long term platform partnerships, and the full picture is at happyporch.com/services.