Google Search Central published a new episode of The Non-Canonical Cabinet on October 14, 2026, covering which characters are allowed in domain names and URL paths, how Punycode and percent encoding work, and why keywords in URLs barely matter for ranking.
Domain names follow internet standards, not Google rules
Domains must comply with RFC 1035 and RFC 1123, which allow ASCII letters, numbers, and hyphens. These rules come from ICANN and the IETF, and Google supports whatever works on standard web infrastructure. Watch from 3:46
Internationalized domain names (IDNs) let you use Chinese, Arabic, and other non-Latin scripts. Behind the scenes, they convert to Punycode, an ASCII string that begins with the "xn" prefix, while browsers usually display the native characters to local users. Google notes this can help brand recognition for local businesses.
Google also warns against lookalike characters. Using an international character that resembles a Latin letter, such as a Cyrillic character that looks like an English "p," confuses users and browsers alike.
Emojis are not allowed in domain names
Although an emoji would technically convert to Punycode, internet standards for domains do not permit them in domain names. Watch from 5:18
Keywords in URLs are an incredibly small ranking signal
According to Gary Illyes, URL keywords only matter when Google has nothing else to go on, such as a brand new site with no content. Once a page has content, keywords in the domain and URL become basically irrelevant. Watch from 6:04
URL paths allow far more characters
Under Internet Standard 66 (RFC 3986), URL paths can contain UTF-8 characters, including non-English text, emojis, and special characters such as exclamation marks and parentheses. Googlebot crawls and processes these URLs.
The key distinction is encoding. Domains are limited by older DNS system constraints, so they rely on Punycode. Everything after the domain uses percent encoding, which converts a character's UTF-8 value into percent signs and hexadecimal values. Watch from 8:26
| Where it appears | Example |
|---|---|
| What you type | example.com/🐔-gear |
| What the server receives | example.com/%F0%9F%90%94-gear |
| How some apps display it | Encoded percent string, emoji, or garbled text depending on the platform |
Google's point is that the path works technically, but a link that looks broken in a social preview or chat message may not get clicked.
Google treats underscores as word joiners
In URL paths, Google reads words linked with underscores as one long word. Google recommends hyphens because they are the universal word separator across the web and are easier for both people and systems to read. Watch from 10:42
Should you change your existing URLs?
Google did not address URL migrations in this episode, so this is our guidance: not necessarily. Underscores or unusual characters in established URLs are not automatically a ranking problem. Changing URLs introduces 301 redirects, broken internal links, and canonicalization risks, so weigh the actual benefit first. For new URLs, use lowercase words separated by hyphens.
When auditing URLs, check for:
- Underscores used as word separators
- Uppercase letters that create duplicate URLs
- Inconsistent UTF-8 encoding in non-ASCII paths
- Double-encoded characters
- Lookalike characters in internationalized domains
For a full walkthrough, see SEO-Friendly URLs: How to Structure and Audit URLs .
The bottom line
Google's recap comes down to three rules: keep domains standard, know that paths can technically handle almost anything, and choose simple, readable URLs with descriptive words and hyphens anyway. Stuffing keywords into URLs to game rankings does not work.
For Google's documentation, see URL structure best practices .