TL;DR
- Google Search Central announced 'The Non-Canonical Cabinet' with a teaser on 6 Oct 2026; episode one, 'What can you actually put in a URL?', went up on 14 Oct 2026.
- Gary Illyes said URL keywords only matter when Google has 'absolutely nothing else to go on', and become 'basically irrelevant' once content is seen.
- Google treats underscore-joined words in a path 'as one long word' and recommends hyphens, matching its URL structure documentation.
- Emoji are not allowed in domain names (2017 ICANN Board resolution) but work in URL paths, where a chicken emoji is percent-encoded as %F0%9F%90%94.
- Search Engine Roundtable notes the advice repeats years of Google statements; Google says the next episode covers robots.txt, and Search Engine Journal notes no release schedule was given.
Google Search Central published the first full episode of a new YouTube series, "The Non-Canonical Cabinet", on 14 October 2026. In the episode, titled "What can you actually put in a URL?", Google Search Developer Advocates Gary Illyes and Cherry Prommawin explain what a domain name and a URL path may contain, and Illyes calls keywords in a URL "an incredibly small signal" that matters only when Google has nothing else to go on. Most of the advice repeats guidance Google has given for years. What is new is the format and the fact that the statements are now on record in one place, as Search Engine Journal reported the same day.
What the Non-Canonical Cabinet series is
Google Search Central announced the Non-Canonical Cabinet with a teaser video on 6 October 2026, according to Search Engine Journal. The trailer describes it as a sitcom-style comedy built on specific questions asked by website owners and the SEO industry. The episode description puts the scope this way: "These are questions that don't belong on a standard FAQ page but are still useful for Search and generative AI features."
Search Engine Journal points out that Illyes also hosted "How Search Works", a five-part Search Central series from 2024 that walked through crawling, indexing and serving in order. The new series runs the other way round. It starts from the odd question a site owner actually asked and works back to the technical rule behind it. Google has asked viewers to post their own edge-case questions in the comments under each video, so the topic list is partly crowd-sourced.
The trailer and the first episode went up eight days apart. Search Engine Journal reports that the next episode covers robots.txt and that neither video description gives a release schedule.
Domain names and URL paths follow different rules
The 14 October 2026 episode separates two parts of a web address that people often lump together: the domain name (the host, such as example.com) and the path (everything after the first slash). The rules for each are set by different bodies and are much stricter for the domain.
Domain names: letters, numbers and hyphens
Prommawin said in the episode that Google follows the technical standards set by the Internet Engineering Task Force for domain names. Search Engine Journal names those standards as RFC 1035 and RFC 1123, which allow letters, numbers and hyphens in a domain name. ICANN also sets policy at this level.
Internationalized domain names and Punycode
Domains written in non-Latin scripts, such as Chinese or Arabic, are called internationalized domain names (IDNs). Illyes and Prommawin explain that these are converted into Punycode, an ASCII string that begins with "xn--". Browsers usually show the IDN in its native script, but the underlying DNS name is the xn-- form. Search Engine Roundtable's topic list includes "why 'xn--' is not a typo", which is a fair summary of the confusion this causes in reports and logs.
Illyes also flagged a known weakness of IDNs. Some scripts contain characters that look almost identical to characters in other scripts. Search Engine Journal gives the example of a Cyrillic letter that is very close in appearance to the Latin "p". Lookalikes like this can confuse both people and browsers.
No emoji in domain names
Illyes said current internet standards for domains do not allow emoji in the domain name. Search Engine Journal backs this up with a 2017 ICANN Board resolution stating that emoji are "disallowed by the current technical standards for Internationalized Domain Names (IDNs)." The same resolution notes that some country-code top-level domains still let people register emoji at the second level. Search Engine Roundtable's summary of the episode puts it more bluntly: a chicken-and-fire emoji domain "can't" exist.
URL paths: emoji and non-Latin text are allowed
The path is far more permissive. Illyes said Googlebot will crawl a path that contains an emoji and "our system will actually process it." Browsers percent-encode such characters before sending the request, so the chicken emoji used in the episode becomes %F0%9F%90%94 in the path. Illyes added that not every browser or social media app will render the emoji back, and some display "a string of gibberish" instead.
None of this contradicts Google's written documentation. Google's "URL structure best practices for Google Search" page says Google Search supports URLs as defined by IETF STD 66, recommends percent encoding for characters in the non-ASCII range when linking, and already uses an emoji path as one of its own examples.
Keywords in URLs: "an incredibly small signal"
The line most SEO teams will quote from the episode is about keywords. Illyes called keywords in a URL "an incredibly small signal" and said, "It really only matters when we have absolutely nothing else to go on." His example was a site that is seconds old and has no content. Once Google can see content, he said, keywords in the domain and URL "become basically irrelevant."
Search Engine Roundtable's Barry Schwartz adds that Illyes made a point of saying Google can detect spamming in URLs. Search Engine Roundtable's topic list jokes about whether a URL like "best-grilled-chicken-cheap-delicious-amaze-amaze-amaze" will make a page number one forever, and Search Engine Journal reports that the closing recap tells viewers to stop trying to game results by packing keywords into URLs.
This is not new guidance
Search Engine Roundtable is explicit that "these are actually all topics we covered before." Its 14 October 2026 post links to more than a dozen earlier write-ups on URLs as a ranking signal, with headlines such as "Google: Keywords In URLs A Very Small Ranking Factor" and "Google Really Wants You To Avoid Changing URLs Just For SEO Reasons." Search Engine Journal cites John Mueller of Google calling URL keywords a "very small ranking factor" in 2016 and a "very very lightweight factor" in 2021. The 2026 wording from Illyes is sharper than either, but the direction has been the same for a decade.
So the honest framing is this: the episode does not change how Google treats URLs. It gives SEOs and developers a recent, first-party clip to point at when a stakeholder asks for keyword-heavy slugs or a site-wide URL rewrite "for SEO".
Hyphens, not underscores
On word separators, Illyes said Google treats words joined by underscores in a URL path "as one long word." He added that this is sometimes fine, because Google may still be able to work out the individual words, but that Google recommends hyphens as the common separator used across the web.
Google's URL structure documentation says the same thing and explains why. It recommends hyphens instead of underscores because hyphens help users and search engines identify concepts in the URL. It does not recommend underscores "for historical reasons": underscores are commonly used to mark concepts that should be kept together, such as function names like format_date in programming languages. The documentation also lists joining words with no separator at all (its example is "greendress") as not recommended.
Search Engine Roundtable's archive on this one topic runs from Matt Cutts-era posts about underscores to a later headline saying "Google: Hyphens Still Recommended Over Underscores In URLs For SEO." The episode settles it in Google's current voice.
Episode claims next to Google's documentation
The table below sets the main statements from the 14 October 2026 episode, as reported by Search Engine Journal and Search Engine Roundtable, against what Google's URL structure documentation already says.
| Topic | What the episode said (14 Oct 2026) | Google URL documentation |
|---|---|---|
| Keywords in URLs | "An incredibly small signal"; "basically irrelevant" once Google sees content | Recommends descriptive, readable words over long ID numbers; no ranking weight stated |
| Word separators | Underscore-joined words are treated "as one long word"; hyphens recommended | Use hyphens, not underscores; avoid joining words with no separator |
| Emoji | Not allowed in domain names; allowed in paths, which Googlebot crawls and processes | Percent-encode non-ASCII characters; an emoji path is one of its examples |
| Non-Latin domains | IDNs are converted to Punycode starting with "xn--" | Not covered on the URL structure page |
What Google did not say
A few things are absent from the sources, and it is worth being precise about them. The episode, as reported, gives no figure for how much a URL keyword is worth, beyond calling it incredibly small. It does not say Google penalises underscore URLs; Illyes said underscores are sometimes fine. It does not tell site owners to change existing URLs. Search Engine Journal draws the practical conclusion that sites already following the URL structure documentation need no changes.
The sources also say nothing specific about any one language or market, Thai included. Anything below about Thai URLs is analysis based on how the standards work, not something Google said in the episode.
What this means for Thai marketers
This section is analysis. Thai is a non-Latin script, so both halves of the episode apply directly to Thai sites.
Thai-script slugs are percent-encoded
A Thai word in a URL path is sent as UTF-8 bytes and percent-encoded, exactly like the emoji example. Each Thai character takes three bytes in UTF-8, so the three-letter word ไทย becomes %E0%B9%84%E0%B8%97%E0%B8%A2 on the wire. That is why a Thai slug pasted into a chat app or spreadsheet sometimes turns into a long string of %E0 codes: it is the same "string of gibberish" display issue Illyes described for emoji. Google's documentation recommends percent encoding non-ASCII characters in href attributes, and it also recommends using words in your audience's language in the URL, with Japanese as one of its examples. For a site whose audience searches in Thai, that guidance points toward readable Thai slugs being acceptable.
Thai domains use Punycode
The Thai-script top-level domain .ไทย is an IDN, so under the hood it is the Punycode label xn--o3cw4h. The same applies to a Thai-script second-level name. The lookalike-character warning Illyes gave is mostly about mixing scripts, which is worth keeping in mind before buying a Thai IDN for brand protection or campaigns.
Thai slug or English slug?
Teams in Bangkok often argue over whether Thai-language pages should use Thai or English slugs. On Illyes's own account, the keyword value of either choice is close to nothing once Google has the page content. The decision is better made on practical grounds: how the link looks when shared, whether the CMS and analytics tools handle encoded paths cleanly, and consistency with the rest of the site. Changing existing URLs purely to add keywords carries redirect risk with almost no upside, which matches the older Google advice Search Engine Roundtable links to.
If you want a second pair of eyes on URL structure, encoding or redirects before a migration, a technical SEO audit is the right place to check it, and broader SEO services in Thailand cover the content and links that actually move rankings. Sites being rebuilt can bake the hyphen and encoding rules in from the start through web design.
FAQ
Do keywords in a URL help Google rankings?
Barely. Gary Illyes called them "an incredibly small signal" on 14 October 2026 and said they only matter when Google has nothing else to go on, such as a brand-new site with no content. Once Google sees the content, he said, URL and domain keywords become "basically irrelevant."
Should I use hyphens or underscores in URLs?
Use hyphens. Illyes said Google treats underscore-joined words "as one long word," and Google's URL structure documentation recommends hyphens over underscores for separating words.
Can I put an emoji in my domain name?
No, not under current standards. Illyes said domain standards do not allow emoji, and a 2017 ICANN Board resolution describes emoji as disallowed for IDNs, although some country-code domains allow them at the second level. Emoji in the URL path are allowed and get percent-encoded.
Are Thai-language URLs OK for Google?
Google's documentation does not single out Thai, but it recommends using words in your audience's language in URLs and percent encoding non-ASCII characters, which is how Thai slugs work. Thai domain names such as .ไทย are converted to Punycode, as the episode explained for all IDNs.
Do I need to change my existing URLs after this episode?
No. Search Engine Journal notes that sites already following Google's URL structure documentation need no changes, and nothing in the episode asks site owners to rewrite URLs.
The Non-Canonical Cabinet is worth following for the robots.txt episode next, and for the comments, where Google is collecting the edge cases it will answer. For a Thai site, the takeaway from episode one is simple: keep URLs readable, separate words with hyphens, let the browser handle encoding, and put the SEO effort into the page itself.
