My website is down — what to check first

A website that will not load is one of four things: a problem on your own computer or connection, a domain problem, a hosting problem, or a fault in the site itself. The first ten minutes are about deciding which, because each is a different phone call and calling the wrong one costs a day.

Check website status first, then work through the first four checks

1. Load it on a phone with wi-fi turned off. Different device, different network, different cache. If it loads, the site is up and the problem is yours — a stale cache, a browser extension, a network fault or an ISP outage. This eliminates a genuinely common false alarm before anyone is called.

An independent website checker or status checker, or a VPN: unavailable for everyone, or just you?

A VPN does the same job as the second device above, and so does an independent website checker or status checker — a service that reports a site’s status as unavailable for everyone, not only for you, without needing a phone at all. Neither tool says why a site is down, only whether the outage is genuinely site-wide, which is still worth knowing before check 2 below.

Ruling out a network fault or an ISP outage first

If step 1 shows the site loading everywhere except one connection, stop there — the fault is a local network or ISP outage, not the website, and none of the checks below apply.

2. Ask whether email on the same domain still works. Send yourself one and have someone outside the business send you one. This is the single most useful question in the whole process. Email and the website usually depend on the same domain and the same DNS but on different servers, so:

• Website down, email down — the domain or its DNS. Registrar problem, expiry, or a DNS change.
• Website down, email fine — the hosting or the site itself. Host problem, server problem, or a fault in the site.

3. Read what the browser actually says. “Server not found” or a DNS error is a naming problem. A connection timeout is a server that is not answering. An error page with a number on it means the server is answering and something inside it failed, which is a different and usually smaller problem — that one is on reading an error page.

4. Check the two dates. When does the domain expire and when was the hosting last paid. An expired domain and a failed hosting payment are the two most common causes of an outage that arrived with no warning, and both are visible in an inbox somebody stopped reading.

What the symptom points at: server, DNS or hosting

Symptom	Most likely cause	Who fixes it
Loads on phone, not on your computer	Local cache, network or extension	You, in five minutes
Site and email both gone	Domain expired, or DNS changed	Registrar account holder
“Server not found”	DNS or domain	Registrar account holder
Times out, email fine	Hosting or server	Host account holder
Error page with a number	The site or its software	Developer
Site loads but shows a stranger’s page	DNS pointing elsewhere	Registrar account holder
Warning about the connection	Certificate, not an outage	Host, usually in minutes

The last row is worth separating out because it is frightening and is usually trivial. A certificate warning means the site is running and the browser is objecting to how, which is on an expired security certificate.

The causes that recur

An unpaid invoice. Hosting or domain renewal on an expired card, with the reminders going to an address nobody monitors — often a former staff member or a supplier’s inbox. This is the most common cause of an outage with no other explanation.

A domain that expired. Distinctive because email stops at the same moment. The recovery window and its costs are on an expired domain.

A change made yesterday. An update, a plugin, a theme, a DNS edit, a migration. If something changed in the preceding fortnight, start there. What to do about it is on a site broken after an update.

A host-side event. Server failure, maintenance, or resource limits exceeded. Check the host’s status page, which most publish and few customers know about.

A suspension. Hosts suspend accounts for non-payment, for resource abuse, and for hosting compromised content. A suspension usually presents as a specific page rather than a timeout, and it is often the first anyone hears that the site was compromised — see a hacked website.

What to have ready before you call

The call goes faster and costs less if you can state four things without looking them up.

• The domain, and who the registrar is
• Who the host is, and the account holder’s name
• When it stopped, and what changed in the fortnight before
• Which of website and email is affected

If you cannot answer the first two, that is the actual finding, and it is more important than this outage. The audit is on who holds your accounts.

What not to do

Do not change DNS settings to see if it helps. DNS changes take time to propagate and a speculative change turns a one-hour problem into a one-day one.

Do not let anybody reset or transfer the domain during the incident. Recovery from a transfer made in a panic is much harder than recovery from an outage.

Do not delete or reinstall. If the site is compromised, the compromised files are the evidence of how, and reinstalling over them usually leaves the way in intact.

Do not buy a second domain. It solves nothing and it fragments the thing customers already have.

While it is down

Two things are worth doing in parallel with the fix, and neither requires technical access.

Update the business listing with a note if the outage will be long, since that is where most people will land next. And check whether any paid advertising is running to the site, because advertising pointing at a dead address spends money on nothing until it is paused.

Preventing the next one

The three that prevent nearly all avoidable outages: renewals on a payment method that will not expire, notifications going to an inbox the business monitors, and someone responsible for updates. Those are the substance of an ongoing arrangement, and what one should contain is on website maintenance cost.

What to do next

Do check 1 and check 2 now, in that order — they take under two minutes and between them they route the problem correctly. The technical scope of who does what afterwards is on web development services.