Why Websites Go Down: Outages Explained Simply
Updated November 16, 2026 · 6 min read
When a huge website goes down, it's easy to imagine a single server catching fire in a basement. The reality is usually more mundane, and more interesting. Modern websites are chains of many services, run by many companies, and a weak link anywhere can take the whole thing offline.
Quick answer
Websites go down for a handful of common reasons: DNS problems (people can't find the site's address), CDN or network failures (the delivery system in front of the site breaks), expired security certificates, bad software updates or configuration changes, overwhelming traffic or DDoS attacks, and failures at the cloud provider hosting the site. Most outages are caused by changes, not attacks, and most are fixed within minutes to hours.
How visiting a website actually works
To understand what breaks, it helps to know the steps behind every page you load:
- Finding the address. Your device asks a DNS server to turn a name like
example.cominto a numeric address. - Reaching the front door. Your request travels across the internet, usually to a CDN (content delivery network) that sits in front of the site.
- Making it secure. Your browser and the server check a security certificate and set up an encrypted connection.
- Doing the work. The site's application servers build the page, often talking to databases and other internal services.
- Sending it back. The page travels back to you, often cached by the CDN for the next visitor.
Each step depends on different systems, often run by different companies. Here's how each can fail.
DNS problems: the address book breaks
If DNS fails, your browser can't find the site at all, even if every server behind it is fine. That's why DNS outages feel so total: the site seems to vanish.
DNS can fail because:
- Someone published a wrong record, pointing the domain at the wrong place.
- The DNS provider itself has an outage, taking many unrelated sites down at once.
- The domain expired. Yes, even big companies occasionally forget to renew.
Visitors see errors like DNS_PROBE_FINISHED_NXDOMAIN. Because DNS answers are cached, a fix can take a while to reach everyone, which is why the same site may work for one person and not another. If you suspect your own DNS rather than the site's, see how to flush DNS.
CDN and network failures: the front door jams
Most big sites sit behind a CDN, a global network of servers that delivers pages quickly and absorbs traffic. When a CDN has a problem, every site using it can go down at the same moment. That's why you sometimes see a cluster of unrelated, popular sites all failing together.
The internet's routing system can fail too. Networks announce to each other which addresses they can reach. If a large network makes a mistake, traffic can be sent to the wrong place or nowhere at all.
You might see Cloudflare-style 52x errors or 502/504 errors. Our guide to HTTP status codes explains what each means.
Expired certificates: the padlock breaks
Every secure website has a certificate that proves it is who it says it is. Certificates expire, deliberately, for security. If nobody renews one in time, browsers show a scary warning and many apps refuse to connect at all.
Automated renewal has made this rarer, but it still happens, particularly for less-visible services like APIs used by mobile apps. The site is technically up, but almost nobody can use it.
Bad updates and configuration changes: the most common cause
Ask the engineers who run big websites, and many will tell you most outages start with a change. Companies update their software constantly, sometimes many times a day. Usually it's fine. Occasionally a change contains a bug, or a small configuration setting is wrong, and it spreads to every server before anyone notices.
Good teams roll out changes gradually and can undo them quickly, which is why these outages often last minutes rather than hours. When you see a 500 error, a recent change is a likely suspect.
Too much traffic: success becomes a problem
Websites are built to handle a certain amount of traffic. Ticket sales, product launches, game releases, breaking news and sports finals can bring far more visitors than usual at once. Servers slow down, then queue requests, then start refusing them with 503 errors or timeouts.
Many sites now use virtual waiting rooms for big events, which is a deliberate trade-off: some people wait so that nobody gets nothing.
DDoS attacks: deliberate flooding
A distributed denial-of-service (DDoS) attack floods a site with junk traffic from many machines at once, trying to drown out real visitors. Large sites use specialised protection services to filter it out, and most attacks are absorbed without anyone noticing. Occasionally one gets through, or the protection blocks real visitors along with the junk.
Attacks get a lot of attention, but they're responsible for a smaller share of outages than most people assume.
Cloud region failures: someone else's data centre
Most websites rent their servers from large cloud providers rather than owning them. Those providers split their infrastructure into regions, clusters of data centres in one geographic area. If a region has a power, cooling, networking or software problem, every site that runs only in that region can go down.
Well-prepared services run in more than one region so they can switch over. Many don't, or have hidden dependencies on a single region, which is why one cloud incident can affect many unrelated apps at once.
Databases and internal services
Behind the scenes, a website is many smaller services talking to each other: logins, payments, search, notifications and more. If a database slows down or a key internal service fails, the homepage might still load while everything you actually want to do fails. That's a partial outage, and it's why our status pages include "down for me too" reports alongside our automatic checks.
Why some people can still use a site during an outage
Outages are rarely all-or-nothing:
- Cached pages from a CDN may keep working while fresh content fails.
- Different regions may be served by different data centres.
- DNS caches mean a fix (or a fault) reaches people at different times.
- Apps and websites may use different servers.
So "it works for my friend" doesn't always mean the problem is yours.
What you can do during an outage
- Confirm it on the Is It Down Everywhere? checker or the site's status page here.
- Check the company's official status page or social accounts for updates.
- Don't reset or reinstall anything. It won't help with a server-side problem.
- Wait, then retry. Most outages are fixed within minutes to a few hours.
If the site is up for us but not for you, the cause is closer to home. Start with is it down for everyone or just me? For games and chat apps, see what to do when a game or app server is down. More answers live in the FAQ.