How to check a sitemap for errors

Request the sitemap URL and confirm a 200 status with an XML content type, no redirect chain, and no authentication. Parse it as XML: a single urlset root with url children, each containing loc and optionally lastmod, changefreq, and priority, all in the sitemaps.org namespace. For a sitemap index, a sitemapindex root with sitemap children each containing loc. Compressed sitemaps must be valid gzip. A byte-order mark, an unescaped ampersand in a URL, or a stray character before the XML declaration will all break parsing.

Each sitemap file may contain at most 50,000 URLs and be at most 50 MB uncompressed. Larger sites use a sitemap index pointing at multiple files. URLs must be absolute, use the same scheme and host as the sitemap location, and be percent-encoded correctly. A sitemap can only include URLs on the host it is served from, or on hosts that reference it from their robots.txt.

Fetch every URL in the sitemap, or a large sample on big sites, and check three things. Status must be 200; redirects and 404s should be removed or replaced with the final URL. The page's canonical must point to itself; a URL whose canonical points elsewhere is telling the engine to index a different page and should not be listed. The page must be indexable; a noindex tag or header contradicts the sitemap. Programmatic sites should also confirm that draft, preview, paused, and deleted pages never appear, which is why SerionFlow keeps only live pages in its generated sitemap and removes paused or deleted pages immediately.

Google uses lastmod to decide what to recrawl, and it ignores lastmod on sites where the values are obviously untrue, such as every URL sharing today's date on every fetch. Set lastmod only when the page content actually changed, in W3C datetime format. changefreq and priority are ignored by Google and can be omitted to save bytes.

After submitting, Search Console's sitemap report shows Success, Has errors, or Couldn't fetch, with the number of discovered URLs. Errors include parse failures, invalid URLs, and unsupported formats. "Discovered" is how many URLs were read, not how many were indexed; compare with the Pages report to see how many are actually indexed and why the rest are not. Bing Webmaster Tools shows a similar report and additionally lists individual URLs it could not process. Neither report checks canonical or noindex conflicts, which is why pass three is done by you.

Run the four passes monthly and after any deploy that changes routing or page status. A free sitemap checker handles the fetch, parse, and structure passes in one go; SerionFlow's free sitemap checker does this for any URL without an account. For URL hygiene at scale, a crawler pointed at the sitemap URLs is the practical tool.