What actually happens in the first hour your website goes down

What actually knocks a site offline

A customer texts a screenshot: a white screen, or "This site can't be reached," or a red warning about an expired certificate. That's the whole diagnostic picture you start with, and it's enough to work from.

Sites go down for a short list of reasons. A hosting server has a fault or gets overloaded. A domain or SSL certificate lapsed because a renewal email got buried in a junk folder. An update broke something on the last deploy. Someone made a bad edit to a config file by hand. Occasionally a host suspends an account over a billing issue nobody flagged in time.

A lapsed certificate is a good example of how misleading this can look. The site itself is completely fine underneath. Only the certificate expired, but browsers block the whole page anyway, with a warning that looks far scarier than the actual problem.

Some of these are minutes-long blips on the host's side. Others sit there broken until a person fixes them by hand. You can't always tell which from the outside, and that's fine — the next step tells you.

None of these need you to know code. They need you to know which bucket the outage falls into, because that decides who you call first and how urgent this actually is.

Is it your site, or is it everyone's

Before anything else, check whether the problem is really yours. Load the site from your phone on cellular data, not the office wifi — a local network issue can look exactly like a dead site. Try a different browser. Ask whoever reported it if someone on a different connection sees the same thing.

Check one more page too, not just the homepage. A single broken page can look like a full outage from where you're sitting, especially if that page happens to be the one you check first.

If it loads for you but not for them, the fault might sit between their device and the internet, not your server. If it's down everywhere, for everyone, on every device, that's a real outage and it's time to start working the list.

A quick outage-checker site will confirm in seconds whether the problem is server-side or local to one visitor. It's a free step. Run it before you call anyone, so you're not troubleshooting a problem that isn't there.

Who to call, and in what order

Start with whoever holds the hosting account, because that's where most outages actually live. If a designer or agency manages hosting for you, they're the first call. If you hold the account yourself, log in directly and check the host's status page before doing anything else — many hosts post known outages there in real time, before you'd ever hear about it any other way.

Having that first contact written down somewhere before an outage happens saves real time. A frantic search for an old email thread, at the exact moment you least want to be searching, is not how you want to spend the first ten minutes of a down site.

Keep the list short: one name for hosting, one for the site itself, and how to reach each of them outside of email if it's urgent. A phone number beats an inbox when the clock is running.

Only after ruling out the host does a mistake on the site's own code become the likely cause — a bad update, a broken plugin, a change that didn't get tested first. Whose job it actually is to fix a break depends on what you agreed to before this happened, not on who picks up the phone fastest today.

An alert that warns you, or silence that doesn't

Some setups catch this before a customer ever does. A monitoring tool pings the site on a short interval, often every few minutes, and sends an email or text the moment it stops responding. Someone can be fixing the problem before a single visitor notices anything is wrong.

Other setups have nothing in place at all. The first sign of trouble is a complaint, a missed booking, or a customer walking away confused and never saying why.

The gap between those two setups is the difference between fixing something in minutes and not knowing it's broken until someone tells you, well after the fact. Hours can pass. Sometimes days. If you don't know which setup you have, that's worth finding out today, not during the next outage.

A basic monitoring tool doesn't need to be expensive or complicated. Even a free one that just checks the homepage every few minutes closes most of this gap.

Why restoring from backup is often the actual fix

Once someone identifies the cause, the fix is frequently not a delicate repair. It's rolling the site back to the last known-good version from a backup, taken before the bad update or the broken edit happened. That's faster and more reliable than trying to hand-patch whatever broke, piece by piece, under pressure.

A restore isn't magic, though. It brings back the site as it existed at the moment of that backup, nothing more. Anything changed or added since then — a new blog post, an edited price, a form submission — doesn't automatically come back with it.

This only works if a backup exists, if it was taken recently enough to matter, and if someone actually has the access to restore it. Whether your site has backups, and who can pull the trigger on restoring one is worth knowing before you need the answer under pressure.

Write down what happened while it's fresh

Once the site is back, the outage isn't fully closed out until someone notes what caused it, who fixed it, and how long it took from first report to resolution. Keep it short. A few lines is enough.

That note is what tells you, next time, whether this is a pattern or a one-off. One outage in a year is not the same story as three outages in a month, and you'll only know the difference if something got written down each time it happened.

It also matters if you're paying anyone for ongoing upkeep. Where downtime response fits inside the rest of what keeps a site running is worth understanding on a calm day, not while you're refreshing a broken page.

More from the blog

All posts