44 domains account for 48.6% of 622.7 million subdomains

These figures describe the release of 20 September 2026. The current product release is newer; see the store for current availability and contents.

With a list of 622.7 million subdomain names, the first useful question is not “how many?” but “how are they spread?”. The answer is: very unevenly. A handful of registrable domains carry an enormous share of the names, most domains that have subdomains in this file have just one, and a few naming habits repeat millions of times.

Among domains with subdomains, most have just one

The subdomain file contains names below 120,853,104 registrable domains. Among those domains, 92,853,426 (76.8%) have exactly one recorded subdomain. The median is one; the mean is 5.15, pulled up by the long tail below.

A few domains carry millions

At the other end, 44 registrable domains each have more than one million subdomains in the file. Together they account for 302,560,941 rows — 48.6% of all subdomains. The largest single domain accounts for 66,037,503.

If you analyze this file, group rows by registrable domain before interpreting raw totals. Weighting does not make the file a representative inventory of the internet.

Large shared namespaces in this file

Some selected namespaces contain many rows. The counts below group subdomain names by registrable domain; they do not identify who created, owns or operates each name.

These eight namespaces hold 79,689,423 rows, 12.8% of the file. The counts vary widely between namespaces: in this release github.io has 12 rows and blogspot.com has 19. They are counts in this snapshot, not a measure of any platform’s size, usage or market share.

Common labels in subdomain names

Leftmost label — the first label, such as www in www.example.com:

• www: 127,418,433 names, 20.5% of the file
• mail: 8,448,491 · webmail: 5,258,482
• api: 2,812,838

Labels anywhere in the subdomain part — grouped into categories that can overlap:

• Mail labels (mail, webmail, smtp, autodiscover, autoconfig and similar): 22.6 million names (3.6%)
• Labels often used for hosting-control-panel naming (cpanel, webdisk, cpcalendars, cpcontacts, whm): 19.4 million (3.1%)
• Development and test labels (dev, staging, test, qa, uat, demo, beta, sandbox and similar): 23.5 million (3.8%)
• api anywhere: 6.6 million (1.1%)

Across all eight tracked pattern categories — these plus www, admin/login, CDN/static and remote-access labels — 33.5% of rows contain at least one matching label. They are name patterns, not confirmed services: a name that starts with mail. may never have handled email.

How deep the names go

Depth counts the labels to the left of the registrable domain (a.example.co.uk is depth 1).

The average depth is 2.6 labels. The deepest name in the file has 120 labels.

What this data cannot tell you

• Whether a name is live, resolves in DNS, or serves a website today.
• When a domain was registered or first seen anywhere.
• Who owns a name, where it is hosted, what language or country it serves.
• Everything on the internet. Coverage is incomplete; differences among the selected namespaces are counts in this release, not a measure of all names in those platforms.

Try it yourself

Try the format first with the free 1,000-name samples. Both full files are $9 once, with 30 days to download.

Counts for .com, .io, .ai, .dev and other endings are on the TLD pages.