URL Structure
URL structure is how a site organizes its addresses: the folders, words, and hyphens that make up the path after the domain. A clear, hierarchical URL tells a visitor, a search crawler, or an AI system what a page is about before anything on it has loaded. It is one of the few signals that is readable by a human eye and a machine parser at the same time, which is exactly why it still matters in an AI-search world.
Why URL structure matters
A URL is often the first piece of information anyone or anything encounters about a page. /guide/heating/heat-pump tells you the topic and where it sits in a site's hierarchy before a single word of body copy renders. That legibility builds trust with people, who can predict what a link leads to, and it helps crawlers group related pages into topics. It is worth being precise about what this buys you, though: Google has been explicit that no special markup or file format is required for a page to appear in AI Overviews or AI Mode, and the same is true of URL design – a tidy path is not a ranking hack. What it does is remove friction. A page that a crawler can parse and cluster correctly is a page that is easier to surface correctly, whether the consumer is a person, a search engine, or a language model summarizing the web.
How a good URL structure works
A clean URL follows a handful of habits. Keep it short and use descriptive words instead of IDs or abbreviations. Separate words with hyphens, never underscores or spaces. Let folders mirror the content hierarchy – main topic, subtopic, specific page – so the path reads like a table of contents. Strip out session IDs, tracking parameters, and unnecessary query strings. Once a URL is published, treat it as close to permanent: every later change means a redirect, and redirects are where accumulated visibility leaks away. Lowercase spelling, consistency across the site, and a flat structure without excessive nesting round it out. None of this requires an AI-specific file or extra markup – it is the same technical hygiene that has always underpinned good information architecture.
Common mistakes
Most problems come from legacy systems. Auto-generated addresses like /page?id=48213 tell a visitor or a crawler nothing. Duplicate paths for the same content – with and without a trailing slash, or with different tracking parameters appended – split signals across what should be one page. Nesting five or more folders deep buries content and makes it hard to reach in a few clicks. Unencoded special characters, spaces, or accented letters turn readable words into unreadable percent-encoded strings. And changing a URL without setting up a 301 redirect throws away whatever authority and inbound links that address had built up. None of these are exotic – they are the ordinary debt that accumulates on a site that has been redesigned a few times without cleanup.
Relation to AI visibility
AI assistants such as ChatGPT, Claude, Gemini, and Perplexity crawl web content and frequently cite a source URL alongside their answers. A descriptive path makes that citation more legible and helps a model place the page in the right context. But it is worth keeping expectations realistic: Ahrefs found that only about 6 to 8 percent of URLs ChatGPT cites overlap with a query's Google top-10, and roughly 80 percent of ChatGPT-cited URLs do not rank in Google's top 100 at all – citation by an AI engine is a meaningfully different selection process from ranking in search. What a clean URL structure does contribute is making a site's topic clusters legible to crawlers like GPTBot, so related pages are recognized as belonging together. It is a supporting signal, not a lever that moves citation rate on its own – factors like third-party brand mentions correlate far more strongly with being cited.
Example
Imagine an online shop for bicycles. A bad address reads shop.com/index.php?cat=7&prod=1288&ref=xy. No one can tell what it is about. The tidy version reads shop.com/mountain-bikes/hardtail/trailblazer-29. From the address alone, every person and every AI system can read off: this is about a hardtail mountain bike named Trailblazer in the mountain bikes category. If someone clicks this source in an AI answer, they know immediately where they will land. If the shop later moves the product, it sets up a 301 redirect so the accumulated visibility is preserved.
Common questions
Should I put keywords in the URL?
Yes, but naturally. One or two descriptive terms that name the page content are useful and readable. Stacking many keywords together to game something looks spammy, provides no measurable benefit, and Google has said explicitly that no AI-specific formatting or extra markup improves how a page performs in AI features.
Do I have to keep old URLs, or may I change them?
Changing them is possible, but never without a plan. For every changed address, set up a 301 redirect to the new one. That way links, rankings, and accumulated authority are preserved instead of running into a dead end – and any AI system that had previously cited the old path will follow through to the new one.