Google rarely revisits something as basic as how you build a URL. But on October 14, 2026, Gary Illyes and Cherry Prommawin from Google Search Central did exactly that, in the first episode of a new video series called The Non-Canonical Cabinet. They confirmed several long-standing assumptions about word separators, keyword stuffing, and emoji in URL paths. If you have not run a URL structure SEO audit in a while, this is a good week to do one.
The short version: hyphens still beat underscores, keywords in a URL barely move rankings once Google has content to read, and non-ASCII characters need proper encoding, or they cause more trouble than they are worth. None of this is radical. Most of it restates guidance Google has published for years. What changed is that two Googlers said it out loud, on camera, with specific examples, which gives you a clean reason to check your own site against it.
Why Google Revisited URL Structure This Week
Google Search Central teased this series on October 6, 2026, and published the first episode on October 14, according to Search Engine Journal’s coverage of the launch. The format is a loose conversation between Illyes and Prommawin about edge cases nobody typically asks about: what characters are legal in a domain, how internationalized domain names get encoded, and whether an emoji can survive a trip through a URL path.
Illyes called keywords in a URL or domain “an incredibly small signal” for rankings, a line picked up by both Search Engine Journal and Search Engine Roundtable’s recap of the episode. He said the signal matters most when Google has almost nothing else to evaluate, such as a brand-new domain with no content indexed yet. Once Google can read the actual page, keyword placement in the URL becomes close to irrelevant.
On separators, Illyes repeated guidance that has sat in Google’s documentation for over a decade: use hyphens, not underscores, to break up words in a path. Google’s Search Central documentation on URL structure states plainly that it recommends “using hyphens (-) instead of underscores (_) to separate words in your URLs,” because underscores get treated as part of one continuous word rather than a break between two.
The episode also covered emoji and non-ASCII characters in URL paths. Illyes confirmed Googlebot will crawl a path containing an emoji, but the episode’s own closing message, paraphrased across coverage, comes down to this: just because you can do something does not mean you should. Browsers percent-encode these characters into long strings of letters and numbers, and not every platform renders them back into something readable. A chicken emoji in a path becomes %F0%9F%90%94 once encoded, which is not something you want showing up in a shared link or a browser tab.
The URL Structure SEO Audit: 8 Checks To Run
None of these URL structure checks need a developer sprint. Most take fifteen minutes with a crawler like Screaming Frog, a bulk URL export from your CMS, or Search Console’s URL inspection tool. Work through them in order, since a few build on each other.
1. Confirm You’re Using Hyphens, Not Underscores
Pull a list of your URLs and search for underscore characters in the path. If you find them, especially on category or product pages, you are asking Google to read two or three words as one unbroken string. That weakens the small amount of signal a readable URL path still carries.
Example: a listing at /product_blue_running_shoes reads to Google closer to “productbluerunningshoes” than to three distinct words. Rewriting it to /product-blue-running-shoes gives you a visible, human-readable path and removes a point of friction.
What to do:
- Export every indexed URL and flag any path segment containing an underscore.
- Fix new URLs going forward, even if you decide not to redirect old ones in a low-traffic section.
- Add a 301 redirect only where the page gets meaningful traffic or links; redirecting an entire low-value legacy section rarely justifies the engineering time.
2. Stop Over-Optimizing Keywords In The Path
If your URLs read like a meta description, for example /best-affordable-waterproof-running-shoes-for-women-2026-review, you are stuffing a field that Illyes says carries an incredibly small amount of weight. The extra keywords add length and clutter without measurable upside once your page itself has content.
A cleaner version, such as /womens-waterproof-running-shoes, keeps a relevant word or two for human readability and leaves the real optimization work to your title tag, headings, and body content, where it belongs.
What to do:
- Cap new product and article slugs at three to five meaningful words.
- Leave existing keyword-stuffed URLs alone if they already rank and get links; a redirect here usually costs more than it fixes.
- Put the effort you would have spent on the URL into the H1 and opening paragraph instead.
3. Check Non-ASCII Characters And Emoji Encoding
Run a crawl and look for raw emoji, accented characters, or non-Latin script sitting unencoded in live URLs. Google’s documentation calls for percent-encoding any character outside the standard ASCII set, following the IETF’s STD 66 specification.
A German product page at /möbel/küche should render in links as /m%C3%B6bel/k%C3%BCche. If your CMS does not do that automatically, some browsers, crawlers, and social tools can mishandle the raw characters, which turns a clean URL into a broken preview or a 404 when someone copies the link somewhere unexpected.
What to do:
- Spot-check international and emoji-containing URLs by pasting them into a plain text editor to see the raw encoding.
- Confirm your CMS encodes automatically; most modern platforms do, but older custom builds sometimes do not.
- Avoid adding emoji to new URLs by hand, even though Illyes confirmed Googlebot can technically crawl them.
4. Trim Parameter Bloat And Session IDs
Open Search Console’s URL inspection tool or a server log and look for URLs loaded with tracking parameters, session IDs, or sort and filter combinations. Google’s own guidance says to keep parameters “to a minimum” and specifically flags referral tags, sorting parameters, and session identifiers as ones that should not create separate indexable URLs.
A single product page spawning dozens of indexed variants, such as ?sort=price&session=8831&ref=email, spreads crawl budget across near-duplicate pages instead of consolidating signal on one canonical version. Parameter bloat is one of the fastest ways to quietly undermine an otherwise solid URL structure.
What to do:
- Canonicalize parameter-driven duplicates back to the clean base URL.
- Move session identifiers into cookies instead of the URL string, as Google’s documentation recommends.
- Block low-value parameter combinations, like infinite filter stacking, in robots.txt rather than letting Googlebot crawl every permutation.
5. Standardize Case Sensitivity
Google treats /Shoes and /shoes as two separate URLs. If your server or CMS does not enforce one consistent case, you can end up with duplicate content spread across case variants, each holding a fraction of the links and signaling the single correct version should carry.
What to do:
- Pick lowercase as your standard, the near-universal convention, and enforce it at the server or CMS level.
- Redirect any mixed-case variant to the lowercase canonical.
- Check your XML sitemap for accidental case inconsistencies, since these quietly create duplicate entries over time.
6. Remove Fragments From Dynamic Content URLs
If a hash fragment (#) changes what a visitor sees, such as a tab or a filtered view, Google’s documentation is explicit: do not rely on fragments to serve different content, because crawlers generally treat everything after the # as part of the same page.
What to do:
- Audit any JavaScript-driven tabs, filters, or single-page views that use # in the URL.
- Move content-changing logic to the History API so each state gets its own crawlable, indexable URL.
- Reserve fragments for their original purpose: jumping to an anchor within one static page.
7. Make Paths Descriptive, Not Just IDs
A URL like /p/48213/ tells a visitor and a crawler almost nothing. Google’s own example contrasts this with a readable path such as /wiki/Aviation, which signals topic and relevance before anyone clicks.
What to do:
- Replace opaque numeric IDs with a short, descriptive slug wherever your CMS allows it.
- Keep the numeric ID in the database if your system depends on it; just do not expose it as the only readable part of the public URL.
- Avoid deeply nested folder structures. Two or three levels is usually enough to stay organized without padding the path.
8. Watch For Spam-Pattern And Near-Duplicate URLs
Illyes noted that Google can detect spam patterns in URLs, which typically means excessive repeated keywords, long strings of parameters, or auto-generated paths built to spin up thousands of thin, near-identical pages.
For example, and this one is a hypothetical rather than a real case: an auto-generated city-and-service site that produces ten thousand URLs like /plumber-in-city-name-near-me-fast-cheap-24-7 for every town in a country is the kind of pattern Google’s systems are built to flag, not reward.
What to do:
- Review any programmatically generated URL sets for templated, keyword-repetitive patterns.
- Consolidate thin, near-duplicate location or variant pages into fewer, genuinely useful ones.
- If you run a legitimate large-scale site, such as a classifieds or marketplace platform, make sure each generated page carries enough unique content to earn its own URL and index entry.
What This Guidance Does Not Change
A few things are worth separating from the headline. First, this is not a new ranking factor announcement. Illyes and Prommawin restated existing documentation with more detail and a few entertaining edge cases. They did not introduce a new signal or a new penalty.
Second, fixing your URLs will not move rankings on its own. If your content is thin or your site has bigger technical problems, cleaning up hyphens and parameters is maintenance, not a growth lever. Treat this URL structure SEO audit as hygiene, not a shortcut.
Third, migrating an entire site’s URL structure to chase a marginal signal is rarely worth the risk. A full URL migration means redirects, possible temporary ranking volatility, and real engineering hours. Reserve that work for sites with a genuine structural problem, such as duplicate content from parameters, broken internationalization, or a CMS that cannot produce readable paths at all, rather than for cosmetic cleanup of a handful of old underscore-separated slugs.
In my experience, teams that chase every small-signal announcement end up spending redirect budget on changes nobody will notice, while bigger crawl and content problems go untouched. Prioritize the checks above by how much actual duplication or crawl waste they are causing, not by how recently Google happened to mention them.
Putting The URL Structure Audit Into Practice
A full site crawl, not a handful of hand-checked pages, is usually what surfaces hyphen, parameter, and encoding problems across an entire domain.
Run the eight-point URL structure SEO audit above against a full export rather than a handful of pages you check by hand. A site of any real size has inconsistencies hiding in old sections nobody has touched in years: legacy blog categories, discontinued product lines, or a staging subdomain that got indexed by accident. A bulk export from Screaming Frog, a batch of Search Console URL inspections, or even a spreadsheet pulled straight from your CMS database surfaces the same patterns far faster than manual spot-checks.
If you want more context on how Google’s crawling and indexing pipeline actually processes what it finds, our breakdown of how long Google indexing actually takes walks through the same crawl-to-serve path these URL checks feed into.
FAQ: URL Structure SEO Audit
Do I need to redirect every underscore URL immediately?
No. Redirect pages that get real traffic or backlinks first. Low-value legacy pages can wait, or get folded into a broader migration later.
Will fixing my URL structure improve rankings on its own?
Unlikely by itself. Illyes described keyword placement in URLs as a very small signal once Google can read your content. Treat a URL structure SEO audit as hygiene that supports other SEO work, not a ranking lever by itself.
Are emojis in URLs actually bad for SEO?
Google confirmed Googlebot can crawl them, so there is no direct ranking penalty. The practical risk is rendering and encoding problems on other platforms and browsers, not a search visibility issue.
How does URL structure connect to other technical SEO work on my site?
URL structure sits alongside crawling, indexing, and rendering. If you have not reviewed related technical areas recently, our guides on PDF SEO rules Google follows and JavaScript rendering issues cover adjacent ground worth checking in the same pass.
Why This Matters And What To Do Next
Google rarely explains this level of URL mechanics in public, so when it does, treat it as a prompt to check your own site rather than a reason to panic. None of the eight checks above demand urgent action, but a few likely apply to any site that has grown organically over several years. A clean URL structure costs little to maintain and removes one more excuse for Google to misread your site.
- Run a full crawl this week, and flag underscores, raw non-ASCII characters, and parameter-heavy URLs.
- Fix the pattern in your CMS or template so new URLs stop recreating the same problem.
- Redirect only the URLs with real traffic or links; leave low-value legacy paths alone unless you are already doing a broader cleanup.
- Recheck in Search Console’s URL inspection tool after changes to confirm Google is reading the corrected paths.
Leave a response