5xx server errors are the family of HTTP status codes from 500 to 599, all meaning the server failed to deliver a page it was asked for. The request was valid; the fault lies with the website, its hosting or something in between, not with the visitor.
How 5xx server errors work
The four you will meet most often each point to a different kind of failure:
- 500 Internal Server Error. The site’s code crashed: a plugin conflict, a PHP error after an update, or a broken configuration file.
- 502 Bad Gateway. A proxy or CDN in front of the site asked the origin server for the page and got an invalid reply.
- 503 Service Unavailable. The server is overloaded or deliberately offline for maintenance.
- 504 Gateway Timeout. The origin server took too long to answer and the proxy gave up waiting.
Search engines react to these codes in a predictable way. When Googlebot starts receiving 5xx responses, it slows its crawling so it does not make the strain worse. If the errors persist, Google begins dropping the affected URLs from its index. Google treats an unreachable robots.txt file returning a server error as a reason to hold back crawling of the whole site until it can read the file again.
For planned work, Google’s advice is to return a 503 with a Retry-After header, which says “back soon” without implying the content has gone. Even then, serving a 503 for more than a day or two risks pages starting to fall out of the index.
Why it matters
An outage has an obvious cost: nobody can buy, book or enquire while it lasts, and any paid clicks land on an error. The less obvious cost comes from intermittent errors. A site that fails for a few minutes at busy times, during a January sale, a Black Friday promotion or the morning an email newsletter goes out, may look fine whenever you check it, while customers and crawlers hit errors repeatedly.
Repeated server errors also slow how quickly Google discovers and refreshes your pages, because it reduces its crawl rate to protect your server. On a large shop that can mean new products or price changes take noticeably longer to appear. See crawl budget for why that matters on bigger sites.
Common mistakes
- Maintenance pages that return 200 or 500. A 200 can get the holding page indexed in place of your content; a 500 says something is broken rather than paused. Use 503.
- Nobody watching. Without uptime monitoring, owners learn about outages from customers, or not at all.
- Error pages cached at the CDN. A brief origin failure gets stored and served to everyone for hours.
- Hosting that cannot cope with peaks. Cheap shared hosting is fine for a small brochure site, but struggles when a campaign or a busy season brings a surge of visitors.
- Blaming an algorithm update. A ranking drop that coincides with server errors is usually the errors.
How to act on it
Set up uptime monitoring that checks key pages every few minutes and alerts you by email or text. In Google Search Console, open Settings, then Crawl stats, and look at host status for server connectivity problems; the Page indexing report lists URLs that returned “Server error (5xx)”. Your host’s error logs show the cause behind each failure.
Fix the root cause: a faulty plugin, too little memory, slow database queries, or a hosting plan that no longer fits the site. Caching and a CDN take load off the origin server. For planned maintenance, schedule it for a quiet period and return 503 with Retry-After. Diagnosing server and crawl problems is part of my technical SEO service.