Infrastructure
Website Traffic Spike After a TV Mention: What Actually Breaks First

If your business is about to get a mention on national television, stop worrying about the server. On properly configured hosting, a sudden crowd of visitors is the part of the night most likely to go fine. What fails is everything on the website that was built for thirty visitors a day and never once tried with thousands: the checkout, the stock count, the order emails, and the plugin somebody installed on the Thursday to get ready.
I have spent more than twenty years running hosting infrastructure, and surge nights follow a pattern consistent enough to be boring. The spike arrives. The homepage holds. Money leaks out of the pages behind it.
A crowd that arrives in minutes, not over a month
Start with the scale, because it is genuinely unusual. RTÉ said last year's Late Late Toy Show averaged roughly 1.7 million viewers across its live and repeat broadcasts, put its share of the Friday night audience at close to four in five, and counted around 642,000 streams on RTÉ Player from 137 countries. The 1.7 million includes repeats and the +1 channel, so it flatters the live moment a little. It is still an enormous room to be mentioned in.
No small business website will see more than a sliver of that audience. It does not need to. Even a sliver can send more people to your site in twenty minutes than you would normally see in a month, and they arrive together, on phones, mostly in the gap between one segment and the next.
A composite surge night
Consider a composite case, built from a pattern that repeats whenever a small maker gets national airtime. A Leitrim maker of hand-painted wooden toys hears, a fortnight out, that one of their pieces will feature on the Toy Show. They do what most owners would. A banner goes on the homepage, a countdown plugin goes in, a dozen new photos are uploaded straight from the phone, and a pop-up appears offering a discount for email sign-ups. None of it is tried on a copy of the site first, because the site has always worked.
On the night, the homepage stays up. The product page is heavy enough on mobile data that a good share of viewers give up before it finishes loading. Those who reach the basket meet a checkout that crawls, and a few pay for toys that sold out an hour earlier, because stock tracking was never switched on. By the next morning the inbox is full of messages asking whether orders went through, and some of the automatic confirmations never arrived at all.
Nothing in that story is a server crash. Every part of it is preparation.

They will search for you, not type your address
Viewers almost never remember a web address from a television segment. They remember a fragment: the toy, the county, the person who made it. So they search, and whatever Google shows for that half-remembered search is your shop window for the night.
That makes timing a technical issue. Google's own guidance on asking it to recrawl a page says crawling "can take anywhere from a few days to a few weeks", and that requesting a recrawl repeatedly for the same address will not get it crawled any faster. A product page you publish at teatime on the day of the broadcast is very unlikely to be in search results that evening.
So the page viewers need has to be live weeks before, not hours. Name it the way a viewer would describe it, not the way your supplier catalogue does. Put the product name, a clear photo and the price near the top. Link it from the top of your homepage too, because plenty of those vague searches will land there first.
The product page has to survive a phone
Product pages can usually be cached, which is good news. A cached page can still be slow if it is carrying eight-megabyte photos and three embedded widgets. Resize images before upload and keep the page focused on one thing: what the viewer saw, what it costs, and the button to buy it.
The homepage can be cached. The till cannot.
Most of a small business website can be served from a cache. When a visitor asks for your homepage, a properly configured server hands back a ready-made copy instead of rebuilding the page from the database every time. Three thousand people looking at the same toy are, as far as the server is concerned, three thousand requests for the same pre-built file. That is why the homepage holds.
The till is different. WooCommerce's own developer documentation tells site owners to exclude the cart, checkout and My Account pages from page caching, because "these pages need to stay dynamic since they display information specific to the current customer and their cart". Every buyer gets those pages built fresh: their basket, their delivery cost, their payment step. Your busiest pages on the night are exactly the ones no page cache can carry.
We have covered that problem in depth in why a slow checkout is the speed problem your homepage test never shows, so I will keep it short. If you only test the homepage before a big night, you have tested the part least likely to fail.
What a slow till costs
In a study Deloitte ran for Google across 37 European and American brand sites and more than 30 million sessions, published on web.dev, a 0.1 second improvement in mobile load times went with retail conversion rates rising by somewhere around 8%. The detail the headline usually drops is more useful: the gains only appeared when every page in the journey improved, not just the first one. It is one study with its own methodology, and I would not quote the exact figure at your bank manager. The direction is not in doubt.
| Page | Can it be cached? | What happens on the night |
|---|---|---|
| Homepage | Yes | Holds well if it is cached and kept light |
| Product page | Usually | Holds, unless it is weighed down by huge photos and widgets |
| Cart and checkout | No, by design | Built fresh for every buyer, so it slows first |
| Order and enquiry emails | Not a page at all | Depend on email delivery, which no cache touches |
The week before is when most sites get broken
This is the uncomfortable part, and I say it as someone whose job is keeping these sites up. The most common way a small business damages its own big night is by changing the production site in the days before it.
Every addition has a cost. A countdown timer, a pop-up, a social feed widget, a chat bubble: each is another script that every visitor's phone has to download and run, cached page or not. If you want the long version of how those costs stack up, it is in our complete performance guide for business websites. The short version is that the week before a surge is the worst possible time to add weight.
Then there is the update nobody tested. A plugin update pushed straight to production on the Wednesday that quietly conflicts with the payment plugin turns a good night into the call no owner wants to take: a stranger on the phone saying their card was charged and no confirmation came.
The operator's answer
The fix is dull, and it works.
- Freeze changes for the last few days before the date. Anything not done by then waits until after.
- Use a staging environment for anything you must change, verify it there, then deploy it to production.
- Take a backup immediately before any change, so a conflict becomes a rollback measured in minutes instead of an evening spent rebuilding pages.
- Place a real test order from a phone on mobile data, end to end, after the last change.
Sold out at ten o'clock
Stock is a website problem on a surge night, not just a warehouse one. In WooCommerce, a product only stops selling by itself when stock management is switched on for it and the quantity is right. Otherwise it keeps taking orders until somebody changes it by hand, which on a busy evening means refunds and apologies the next morning.
If you do sell out, the product page stops being a sales page and becomes the most-visited page on your site. Decide in advance what it says: when more stock is due, whether you are taking pre-orders, and a way to be told when it is back. If you collect email addresses for that, ask for clear consent and say what you will send.
The inbox nobody checked
Order confirmations, enquiry forms and back-in-stock notices all leave your site as email. None of that is helped by caching, and a surge is a bad moment to discover your form emails have been going to junk for months. The week before, send yourself a test order and a test enquiry, and verify that each email arrives in an inbox rather than a spam folder.

What surge-ready hosting actually looks like
Strip the marketing away and a small business site that will cope with national attention needs a short list of things from whoever hosts it:
- Page caching at server level, so the homepage and product pages go out as ready-made copies rather than being rebuilt for every visitor.
- A fast path for the pages that cannot be cached, so the database work behind every basket and checkout is quick.
- A staging copy you can use without a developer, so changes in the run-up are verified before they reach production.
- Backups taken automatically before changes, so a bad update is a rollback, not a rebuild.
- People who answer, in your time zone, if something does go wrong that evening.
That is the standard. Web60 was built to meet it, and I would rather show the working than claim it. Sites run on Nginx with FastCGI page caching, which is what lets the homepage and product pages hold under a crowd. Redis object caching keeps frequently used database results in memory, so the checkout spends less time waiting on the database. In practice, that is the difference between a buyer in a crowded sitting room reaching the payment step and a buyer giving up halfway through the form.
One-click staging lets you rehearse the changes for the big night on a copy of the site. Pre-update safety snapshots and nightly backups with one-click restore mean a conflict on the Wednesday gets fixed with a rollback. Support is handled by real people based in Ireland, not an outsourced queue. The full stack is laid out in the caching layers and Irish infrastructure behind every Web60 site.
There is no agency in the loop either. The AI website builder produces a complete WordPress site in under a minute from a description of your business, and WordPress itself runs roughly four in ten of all websites, according to W3Techs. When you need a new product page the week before a broadcast, you write it yourself rather than waiting on somebody's change-request queue at an hourly rate. All of the above comes in the €60 a year price, with nothing extra for staging or backups. If you are starting from nothing ahead of a big date, describe the business to the builder one evening and you will have a site to work on before you go to bed.
Where a bigger setup genuinely earns its cost
There is a limit to my side of this argument, and it is worth being plain about it.
Caching does nothing for the till. No host can pre-build a checkout, and no host can make your payment provider's pages, your courier's booking system or your stock count any faster. Server capacity on any platform is finite as well. If a crowd arrives far larger than a site was ever sized for, the uncached pages slow first, on any hosting at any price. When you know the date of your moment, ring your host in advance and tell them. A good one will want to know.
Some businesses genuinely need more than this. A national retailer running a heavily promoted flash sale, expecting thousands of orders inside an hour with its own technical team watching the graphs, is the workload enterprise managed hosts with auto-scaling infrastructure are designed for, and they price it accordingly. That is not a one-person maker with a mention on a Friday night show. It is not most businesses in this country.
Prepare the till, not the shop window
A national mention is one of the best things that can happen to a small business, and a well-hosted website will usually stay standing through it. What decides whether it pays off is quieter work: a product page that is already in Google, a checkout you have actually tried on a phone, stock counts that tell the truth, emails that arrive, and a production site nobody touched that week.
If your moment is still a few weeks away, there is time for every item on that list. Start with the till.
Frequently Asked Questions
Will my website crash if my business is mentioned on TV?
On properly configured hosting with server-level page caching, the homepage and product pages usually hold, because they can be served as ready-made copies. The pages that slow first are the ones that cannot be cached, such as the basket and checkout, which are built fresh for every buyer. A crash is less likely than a slow checkout, a stock count that oversells, or order emails that land in spam, so test those before the date.
How far in advance should I prepare my website for a TV or press feature?
Start at least two to three weeks out if you can. Google says crawling a new page can take anywhere from a few days to a few weeks, so the page viewers will search for needs to be published well before the broadcast. Use the final week for testing only: place a test order, send a test enquiry, check stock levels, and avoid changing the production site in the last few days.
Why does my checkout slow down when my homepage loads fine?
Because the homepage can be cached and the checkout cannot. WooCommerce's own documentation says the cart, checkout and My Account pages must be excluded from page caching because they show information specific to each customer and their basket. Every buyer gets those pages built from scratch, so under a crowd they are the first to slow. Object caching, which keeps frequently used database results in memory, helps those pages respond faster.
Should I tell my hosting provider about a TV appearance?
Yes, if you know the date. Tell them when the broadcast or article goes out, which pages you expect people to visit, and whether you are selling online. A good host will want to know, can confirm that caching is working on your product pages, and will know to keep an eye on your site that evening. If your host has no way to take that message, that tells you something useful too.
What should my product page say if the item sells out?
Decide this before the night, because a sold-out page often becomes the busiest page on the site. Say clearly that it is sold out, when more stock is expected, and whether you are taking pre-orders. Offer a way to be told when it is back, and if you collect email addresses for that, ask for clear consent and say what you will send. Make sure stock management is switched on so the shop actually stops selling at zero.
Sources
Graeme Conkie founded SmartHost in 2020 and has spent years building hosting infrastructure for Irish businesses. He created Web60 after seeing the same problem repeatedly — Irish SMEs paying too much for hosting that underdelivers. He writes about WordPress infrastructure, server security, developer workflows, managed hosting strategy, and the real cost of hosting decisions for Irish business owners.
More by Graeme Conkie →Ready to get your business online?
Describe your business. AI builds your website in 60 seconds.
Build My Website Free →More from the blog
The Chrome Suspicious Site Warning: Your Website Does Not Need to Be Hacked to Get One
Chrome now shows a suspicious site warning on lower-confidence verdicts. Why your business website could get one without being hacked, and how to verify it.
WordPress Activity Log: Who Changed Your Website, and When
A WordPress activity log shows who changed your plugins, settings and users. Why any business site with more than one login needs one, and how to set it up.
