Check this page with an assistantOpens a chat asking it to summarise this article and name the evidence behind each claim.
Claude opens with the prompt on your clipboard: Anthropic does not support prefilled prompts on the web, and we would rather copy it than ship a button that drops it.
The Decision You Make at Fifteen Pages and Regret at Three Hundred
Architecture is almost free to decide when a site is small and genuinely expensive to change once it is not. Most sites never decide at all: the structure emerges from whatever the platform defaulted to and whatever seemed reasonable to whoever added the fortieth page.
The symptoms arrive later and rarely get attributed to the cause. New pages take a long time to be discovered. Sections of the site are crawled rarely. Readers cannot find the adjacent thing they wanted. Categories accumulate until none of them means anything. Every one of those is a structural problem presenting as a content or technical one.
Two things are being decided at once and they are worth separating. Hierarchy is how pages relate and connect, which is a real thing with real consequences. URLs are how that hierarchy is expressed as text, which matters much less than people think but is far harder to change later. Get the first right; get the second good enough and stable.
Click Depth Is the Metric That Matters
Click depth is how many links from the homepage it takes to reach a page. As depth increases, crawl frequency falls and internal link equity thins. Keeping pages you care about within about three clicks is a guideline, and the mechanism behind it is what you should actually act on.
The mechanism is simply that crawlers follow links and have finite budgets. A page reachable only through six navigation steps competes with everything shallower for attention, and loses. It is also, not coincidentally, harder for people to find, so it accumulates fewer references and less engagement, which compounds the same disadvantage from a different direction.
Be sceptical of the specific numbers attached to this in vendor research. Claims that flat architectures rank some exact percentage higher than deep ones circulate widely without inspectable methodology, and the comparison is confounded by the fact that well-run sites tend to be both flatter and better in every other respect. The directional advice holds without the statistic.
Choosing a Shape That Fits the Site
The right structure is the shallowest one that still expresses genuine relationships. Flatness is a means, not a goal: a catalogue of ten thousand products flattened into no hierarchy is unnavigable, and a nine-page brochure site with four nested levels is absurd.
| Site size | Sensible shape | Common mistake |
|---|---|---|
| Under 30 pages | Essentially flat, everything linked from navigation | Inventing categories with one page in them |
| 30 to 300 pages | One level of real sections, with hub pages per topic | Relying on chronological archives as the only route in |
| Catalogue or large content site | Category and subcategory, plus cross-links between siblings | Filters as the only path to deep pages, so nothing is crawlable |
| Multi-location or multi-service | A matrix with real pages at the intersections that deserve one | Generating every combination regardless of substance |
The third row is where most large sites actually fail. If the only way to reach a product is by applying two filters, crawlers may never construct that path, and the page is effectively invisible however good it is. Real anchor links to category and subcategory pages are the fix, and the wider handling of filter combinations is in faceted navigation SEO.
URL Rules That Age Well
Short, lowercase, hyphenated, descriptive, and stable. Mirror the hierarchy where the hierarchy is genuinely meaningful, and stop there. The most valuable property a URL has is that it never needs to change.
- Leave out anything that will expire.
Dates, years, campaign names, version numbers, and the current org chart. A URL containing 2026 obliges you either to republish annually or to look stale, and neither is a choice you want forced.
- One level of folder, usually.
Deep folder nesting in the URL rarely adds clarity and makes every future reorganisation more disruptive. A blog post does not need the category in its path unless the category is genuinely permanent.
- Decide trailing slashes and case once.
Then enforce it with redirects. Inconsistency here creates duplicate URLs quietly and forever, which is the mechanism described in duplicate content.
- Keep them readable out of context.
Someone pasting your URL into a message should be able to guess the page. That is worth more than any residual keyword effect, and it is the reason numeric identifiers make poor URLs even though they work fine technically.
- Do not restructure for tidiness.
Every migration carries risk and the payoff for cosmetic improvement is close to zero. Restructure when the current scheme prevents something you need to do, and when you do, treat it as the migration it is, per website redesign SEO.
Questions People Ask About Site Architecture
- What is site architecture in SEO?
It is how your pages are organised and connected: what links to what, how many clicks separate a page from the homepage, and how URLs express that hierarchy. It determines what gets discovered, how often pages are recrawled, and how readers move between related pages, which makes it upstream of most other on-site work.
- How many clicks deep should a page be?
Keep pages you care about within about three clicks of the homepage. This is a guideline rather than a rule, and the reason behind it matters more than the number: pages that are hard to reach get crawled less often and receive fewer internal links, so they are slower to be discovered and slower to be refreshed.
- Is a flat site structure better than a deep one?
Flatter is generally better for discovery, but the honest answer is that it depends on size. A small site should be nearly flat because there is no reason not to be. A large catalogue needs real hierarchy or its category pages become meaningless lists. The goal is the shallowest structure that still expresses genuine relationships, not flatness for its own sake.
- Do URLs need keywords in them?
A readable URL that describes the page is useful to people deciding whether to click and to anyone reading a link out of context. The direct ranking effect is small and has been for years. Prioritise short, stable, and human-legible over keyword density, and never change a working URL purely to add a keyword.
- Should I restructure my URLs?
Only for a substantial reason, because every change costs you. Redirects dilute nothing if implemented correctly but they add hops, break old references that were never links, and create risk in bulk. Restructure when the current scheme actively prevents growth, not to tidy it. If you do, map every old URL to a new one before you ship anything.
