A page fails with a server error

Why it matters

Google's crawlers slow down in proportion to the number of addresses that return server errors, ignore content sent with a 5xx response, and eventually drop indexed pages that keep failing.

What to do

Open the address in the finding; if it works now, look in your host's error logs for the time of the audit.
If it still fails, ask your developer or host what breaks for this address.
Run a new audit once it answers normally.

What we read

We reuse the page’s recorded responses and its one retry. A final 5xx is confirmed only when at least two attempts were captured and every attempt ended in 5xx. A single 5xx, an unreadable response or a 429 leaves the result unknown.

The homepage has its own check: this rule excludes it when either the requested address or the final address matches the registered site URL. For other pages, one captured page is chosen for each final address. Known internal links, breadcrumb items and sitemap entries explain where the address appears; missing reference details do not prevent a finding for a confirmed server error.

The sitemap status check covers 404 and 410, so a server-error finding stays in this check even when the address appears in a sitemap. This check reuses the existing crawl and adds no requests or billable pages.

Example & evidence

Illustrative example — one page read twice in one audit, not a live audit.

The homepage links to https://bakery.example/order-online:

GET https://bakery.example/order-online
HTTP/1.1 500 Internal Server Error

GET https://bakery.example/order-online
HTTP/1.1 502 Bad Gateway

Both answers are server errors, so the page is reported with the last one: Page repeatedly returned HTTP 502.

If the retry had answered 200 OK, there would be no finding.

Confirm & check again

Check for yourself: open the address a few times in a browser and in a private window, and ask your host for error logs from the time of the audit.

Check again: Full audit — Run another audit. Only an error-free response below 500, other than 429, for the same original address and the same final address can mark this finding as fixed. An unreadable response, leaving the page out of the audit, changing its final address or removing its references does not confirm a fix. Earlier error observations remain in the history, and dismissal stays separate from the technical status.

Limits & sources

The recorded responses show the failure during this audit, not how long it lasted or what Googlebot and visitors received. A cancelled or missing retry cannot confirm a persistent error.

Reference details come only from the captured page chosen for each final address. An address outside the crawl’s page list is not checked here, and its absence does not mark an earlier breadcrumb-related finding as fixed.

Confidence: High because both requests failed with a server error; a single failure is not reported.

Sources

Google, How HTTP status codes affect Google's crawlers — says Google slows crawling in proportion to the addresses returning server errors, and eventually drops indexed pages that keep failing. Checked 25 September 2026.

Editorial review: 4 November 2026

Your homepage fails with a server error — the same test for the homepage, reported as critical.
A referenced page answers with an error — an address the site links to or lists answers a client error such as 404.
robots.txt fails with a server error — robots.txt itself fails with a server error.