Google has found a lighter way to answer the strange technical questions that never quite fit inside formal documentation, and the first outing manages to settle one of the oldest arguments in the field along the way. The Non-Canonical Cabinet premiered on October 14, 2026 as a Search Central video series built around edge case questions, and the opening episode with Gary Illyes and Cherry Prommawin worked through domain names, internationalised domains and Punycode, emoji in domains and paths, percent encoding for non-ASCII characters, and the long running question of whether loading a URL with keywords actually helps a page rank.
On that ranking question, Illyes was direct in a way that leaves very little room for creative interpretation by anyone hoping the opposite answer was coming. He described keywords in a URL as an incredibly small signal, and he explained that the signal really only matters when Google has absolutely nothing else to go on, using the example of a site that is seconds old and carries no content yet. Once Google sees the actual content on the page, he added, keywords in the domain and the URL become basically irrelevant, which aligns with earlier comments from John Mueller in 2016 and 2021 that treated URL keywords as a very small and very lightweight factor.
The separator guidance in the same episode deserves equal attention, because it affects readability and parsing on every new page a team publishes. Illyes explained that Google treats words joined by underscores in a URL path as one long word, and while that is sometimes workable, Google recommends hyphens as the common word separator across the web, a position the URL structure documentation already supports. The episode also covered percent encoding for non-ASCII characters in paths, emoji appearing in one of the worked examples, and a closing recap that told viewers plainly to stop trying to game results by stuffing large amounts of keyword data into a URL.
For a team reviewing an existing site after watching the episode, the practical conclusions separate cleanly into what deserves effort and what does not:
- Existing URLs that are already indexed and performing should stay where they are, because changing an address to chase a signal this small creates redirect work and ranking risk that the possible gain cannot justify.
- New pages should use short, readable paths with hyphens between words, written so a person can predict the content of the page before clicking, and that human test is the standard Google keeps returning to.
- Keyword stuffed slugs that repeat the same phrase in several forms should be treated as a quality problem to avoid on anything new, since Google can recognise that pattern and the episode named it directly.
- International sites need the encoding details handled properly, because Punycode in a domain and percent encoding in a path solve different problems and mixing them up produces addresses that break or confuse.
When a client still pushes for a URL rewrite programme built around keywords, I work through the conversation in a particular order:
- Pull the pages they want to rename and check what those pages earn today, because a performing page carries far more value in its history and links than a keyword in its slug could ever add.
- Show the cost side honestly, including redirect mapping, internal link updates, and the temporary uncertainty that follows any large address change across an established site.
- Redirect the energy toward the content itself, since Illyes anchored the whole discussion on the point that content is what Google uses once it can see the page, and stronger content is where the ranking movement actually lives.
- Agree the house rules for anything new, with readable hyphenated paths and no stuffing, so the question stops resurfacing every time a page is briefed.
This guidance sits comfortably beside the direction Google gave in the recent Google helpful content guidance update, where the through line was also about building pages that satisfy a reader rather than arranging surface signals for an algorithm. URL structure belongs in that same category of craft, because a clean address helps a person trust and share a link, and those human benefits remain real even after the ranking value of the words inside it has been stated this plainly by the person who works on the systems involved.
Failon’s POV: I welcome how blunt this episode is, because the URL keyword debate has survived for years on ambiguity that practitioners could bend toward whichever answer suited a sales pitch. A signal that only matters for a brand new page with no content is not a strategy, and any proposal that leads with renaming established URLs to insert keywords should now be answered with this video. Keep addresses simple, keep them stable, and spend the hours that a rewrite project would have consumed on the content and internal linking that Google actually weighs once a page exists.
Sources: Google Search Central, The Non-Canonical Cabinet premiere episode on YouTube (October 14, 2026), Search Engine Journal coverage by Roger Montti, and Search Engine Roundtable coverage of the same episode.
