Sitemap index or a single sitemap: when to switch

A single sitemap is enough while the site stays under 50,000 URLs and 50 MB uncompressed; past that mark you need a sitemap index, a file that lists the other sitemaps and holds no page URLs itself. The limit belongs to the protocol, not to the server.

The two protocol limits

The protocol caps a file at 50,000 URLs and 50 MB uncompressed. An index accepts up to 50,000 sitemaps, and an index cannot list another index: the hierarchy is two levels deep, not three. gzip cuts the weight, not the URL count.

The shape is <sitemapindex> with one <sitemap><loc> element per child. lastmod is optional at every level, and Google ignores it when it is always today: a value that never changes is worse than no value.

When a single file is the right call

• Under 5,000 URLs an index adds a request and solves nothing.
• When every page changes together, a single lastmod is honest and readable.
• A single file can be checked by hand in a minute, and mistakes show up at once.

When an index earns its place

• Past 50,000 URLs, or when one file exceeds 50 MB uncompressed.
• When sections change at different rates: a daily blog, a monthly catalogue.
• When one section has to be left out without rewriting the whole file.

Read an index before publishing it

cat > check-sitemap.mjs ([^/g)].map((m) => m[1]);
console.log(children.length + " child sitemaps");
for (const url of children) {
  const xml = await (await fetch(url)).text();
  const urls = (xml.match(//g) || []).length;
  if (!urls) console.log("EMPTY   " + url);
  else console.log(String(urls).padStart(6) + "  " + url);
}
EOF
node check-sitemap.mjs

The script fetches the index, lists the children and counts <loc> occurrences in each one. A child with zero URLs is a defect to fix: Google reads an empty file and trusts the rest of the index less. A child with 49,000 URLs is a time bomb, because the next addition breaks the limit.

How the EneUI catalogue does it

The index opens with /sitemap-core.xml, which holds hubs, guides and libraries, then lists the component files. The 1,556 blocks and the 12,808 icons stay in the core; individual records such as sidebar from 21st live in the component files, the ones a single file could never hold. Checked on 1 November 2026.

The limitation

An index does not grow the crawl budget. If the site publishes 111,985 URLs and changes a thousand a day, Google reads a fraction: splitting into files makes the changes readable, it does not persuade a crawler to read everything. The only real lever is to stop publishing pages nobody searches for.

Where to continue

The sequence for getting into the index is in getting a site indexed by Google; how language versions are declared is in hreflang: the common mistakes. If the site base is still undecided, the comparison is in WordPress, Wix or React components.