Pinwheel
I've watched a WordPress store fall over on Black Friday night. Not slowly. A homepage that loads in a second flat on a quiet Tuesday can time out completely within minutes once a big email send lands and the real traffic actually arrives. Good web design in London never announces itself on a quiet Tuesday. You find out what you actually bought on the one night of the year that pays for January, when the site either holds or it doesn't.
These stores often run clean for years, and that's exactly the problem. Nobody stress-tests the plumbing until the whole street flushes at once.
What actually breaks when the traffic hits
A traditional WordPress store is doing an astonishing amount of work for every single visitor. Someone loads a product page, and PHP wakes up, queries a MySQL database, assembles the page from a theme and eleven active plugins, and hands it back. On a normal week that dance happens a few hundred times an hour. During Q4, on the stores I've watched, peak nights run five to ten times a normal week's order volume, so the same choreography has to happen thousands of times a minute.
The database is usually where it snaps first. Every plugin wants a slice of it. Your abandoned-cart tool, your reviews widget, your loyalty scheme, your three separate analytics scripts, all hammering the same connection pool. Add a caching plugin and you have papered over the crack, right up until a logged-in customer with items in their basket bypasses the cache entirely and hits the raw database anyway. That is the exact person you cannot afford to lose. The one holding a full trolley at checkout.
I have watched response times climb from a few hundred milliseconds to ten seconds or more under load. Nobody waits that long for a page. They're gone, and the order goes with them.
What 'headless' means once you strip the marketing off it
The word has been chewed to pulp by vendors, so let me be plain. Headless means you split the shop into two parts that used to be welded together. There is the storefront, the thing customers see and touch, and there is the commerce engine underneath, the part that knows about products, prices, stock, and orders.
We build storefronts on Next.js and run the engine on Medusa.js, an open-source commerce backend you actually own rather than rent. The storefront gets served as pre-built pages from a global network, so a shopper in Leeds gets a page that was assembled once and cached everywhere, not rebuilt from scratch on your server for every click. The engine sits behind an API and only gets bothered when something real happens: an add-to-basket, a stock check, a payment. You can read more about how we structure that on our headless e-commerce service page.
Fashion has nothing to do with it. What matters is that the storefront can absorb a hundred thousand pageviews without the commerce engine breaking a sweat, because ninety percent of those views never reach it.
The Q4 failure modes headless just sidesteps
Three specific things go wrong on peak nights, and a decoupled build removes all three.
The first is the cache-miss cliff I described. When your storefront is static by default, there is no cliff. Traffic scales, cost barely moves, and the person mid-checkout is served by a system built to handle exactly that one job well.
Then there is the plugin cascade. On a monolith, one badly written plugin update in mid-November can take the whole site down, and you will spend Black Friday morning bisecting extensions instead of trading. Separate the storefront from the engine and a front-end change cannot corrupt an order, while an engine update cannot blank the shop.
Cruellest of all is the checkout bottleneck. Your marketing works, traffic pours in, and the single point everyone funnels toward is the one page still talking directly to the database. A proper commerce engine handles checkout as an isolated, queue-tolerant operation. Orders land in order. Nobody double-charges. Stock counts stay honest even when four hundred people grab the last of something in the same second.
A realistic timeline if you start in September
People assume a re-platform means burning the summer. It does not, if you scope it properly.
Starting the first week of September, a focused migration for a small catalogue runs roughly eight to ten weeks. Two weeks to model your products and port data cleanly, three to four to build and style the storefront, two to wire up payments and shipping and test them under simulated load, and a final fortnight of soft launch where the new site takes real orders while you keep the old one warm as a fallback. That lands you live and settled by early November, with a genuine week or two of buffer before the first big sale.
Leave it to October and you are rushing the one part you must never rush, which is testing checkout under pressure. I would rather tell a client to wait until January than push a rushed payment flow into Black Friday.
What to check on your current stack this week
Even if you are not migrating this year, spend an hour on this. Open your site on a throttled connection and time the checkout page specifically, logged in, with something in the basket. That is your real number, not the homepage score a speed tool flatters you with.
Then ask your host one blunt question: what happens at ten times current traffic? If the answer is a shrug or an upsell to a bigger plan, you have found your weak point already. Count your active plugins. Anything over fifteen on a store is a liability waiting for a busy night. And check that your last full backup actually restores, because the worst time to learn it doesn't is at 9pm on the biggest Friday of the year.
If you want a straight assessment of whether your current build will hold, that is the kind of web design in London question worth settling before October, not during it. Talk to us. No pitch, just an honest look at where your Q4 risk actually sits.
The owners who make this move tell us the same thing afterwards, in a different tone entirely: they spent Black Friday evening watching the numbers climb instead of watching a loading spinner. That is the whole game. Building the thing that lets you enjoy your own busiest night.