How to build a functional website architecture and structure

One of the first things I look at in a more comprehensive SEO audit or in ongoing collaboration is website structure and architecture. You can have the best content in the world, but if it is hidden 5 clicks from the homepage, too concentrated or scattered without any system, then Google, users and LLMs will find it late or not at all. Some clients, for example, create high-quality content videos but put them all in one videos section. Instead, they should add them by topic and content to relevant pages that already cover that topic in full. Organising content by topic is always better than organising it by format (at least on a website).

Structure and architecture are how you set priorities. Whatever comes first in the menu is the most important. Full stop.

Website structure is the skeleton that everything else stands on: SEO, UX, conversion paths. When the skeleton is crooked, the whole website limps. Over years of auditing I have seen websites with tens of thousands of URLs where half the pages were effectively invisible because no internal link pointed to them. I have also seen smaller company websites with 20 pages where the structure was well thought out.

In this article I set out how I think about structure: what works, what does not, and how to design it all in practice. No textbook theory, just the process I actually use on client projects. We will go through what website structure actually is, what elements it consists of, what types of structure exist and how to design your own so that it serves SEO, GEO and users at the same time.

What website structure is and why it matters

Website structure is the way pages are organised and linked to each other. It is also called website architecture or information architecture. In practice it is about what the hierarchy looks like: what is on the first floor, what is on the second, how you move between floors and whether you can get to the basement at all.

From an SEO perspective, structure is critical for three reasons:

• Crawlability: Googlebot has to find the page. If no internal link points to it, or it sits at the 5th level of depth, the crawler may miss it entirely. Or it may visit once every two months instead of every day. The closer a page is to the homepage, the more often Google crawls it and the faster it picks up changes. The reverse also applies. Try to push unnecessary pages into the background and bury them properly.
• Authority: every internal link passes ‘strength’ to the target page. A logical structure ensures that the most important pages receive the most internal links. A poor structure scatters that strength across hundreds of irrelevant pages that nobody searches for.
• Indexing: Google understands what a website is about better when it sees a clear hierarchy of topics. Category → subcategory → detail signals that this website has depth in the topic. Flat, unconnected pages, on the other hand, look like a random collection of content without topical authority. Structured data also helps. It divides your pages at a further level and makes them clearer for LLMs and Google.

From a UX perspective it is simple. Visitors want to find what they are looking for within 3 clicks. If they have to think about where to click, they leave. Website structure is essentially a navigation map, and when the map is confusing, people get lost. A lost visitor does not convert.

In audits this is the first thing I deal with. Not technical SEO, not the content on individual pages, but structure. If the structure is wrong, there is no point dealing with the rest. You can have perfectly set titles, a fast website and great content, but if it is all wrapped in a chaotic structure, neither Google nor users will appreciate it.

Lately I keep seeing clients trying to cut back their websites (and with them their digital footprint). The logic is that fewer people now reach the website, so they can trim it, close some sections or regenerate the website in Claude Code and throw away what they spent years building. I have already seen several sad stories of this kind, and I am a big fan of Claude Code.

Types of structure

Not every website needs the same type of structure. In practice I come across three basic models, and most websites I work with use some combination of them.

Tree (hierarchical) structure is the most common and makes the most sense for most websites. Homepage → category → subcategory → detail (with tags or parameters running in parallel). It works for e-shops and larger company websites alike. The advantage is clear logic and scalability. You can add further branches without breaking the rest of the tree. The disadvantage is that if the tree grows too many levels (5+), pages at the ends of the branches become hard to reach and Google gives them minimal attention.

Flat structure means everything is within 1-2 clicks of the homepage. It is typical of smaller websites, portfolios and blogs. The advantage is maximum crawlability, because Google sees everything immediately. The disadvantage is that it stops working once you have more than a few dozen pages. Navigation becomes confusing and there is no topical hierarchy. Without hierarchy, Google does not understand which pages are more important than others. Products, services and the blog are not nested, and the URL does not tell you where you are.

Linear structure is sequential: page A leads to page B, which leads to page C. It is used for onboarding, checkouts and landing pages. It is not a structure for the whole website, but for a specific flow within it. A typical example is a multi-step form or a product selection guide.

Most websites I work with need a tree structure with flat elements: important pages linked directly from the homepage, and the rest in a logical hierarchy. A purely flat website works as long as you have 30 pages. With 300 pages it is chaos. A purely tree-shaped website without shortcuts can be too rigid. Some pages deserve a direct path from the homepage, even if they sit two floors lower in the hierarchy.

What a structure is made of

Website structure is not the menu. It is a set of elements that together form the navigation skeleton. When one element is missing or does not work, the whole structure has a weak spot. Let us go through what it consists of and what to watch out for with each element. Personally, I see structure mainly as a crawl of all the website’s pages, with the content type (page template) of each one labelled.

Main navigation (menu) is the most visible element. A rule that has worked for me is a maximum of 7 items in the main navigation. Every extra item lowers the likelihood that a visitor clicks the right one. On mobile it is even more critical, because nobody scrolls to the end of a hamburger menu with 40 items. I have seen websites where the main menu had over 50 links. The result? The heatmap showed that 80% of clicks went to the first 6 items. The rest was dead space. For B2B websites even fewer items are recommended. According to psychological research, the ideal is just 4 menu items, which the brain can quickly take in to decide where to go. So if you have, say, 6 services, it is best to split them into two main categories, for example, and show those in the menu. I use similar branching on my own website.

Categories and subcategories form the backbone of a tree structure. How many levels are reasonable? A maximum of 3: homepage → category → subcategory → detail. As soon as you add a fourth level, the pages at the end are too far from the homepage and receive minimal internal link juice. That is fine if it is a marginal category for you. However, everything important should be a main category (as always, it depends on the size of the website). Naming also matters. Category names should match what people search for. Not the company’s internal jargon, but real keywords.

Breadcrumb navigation helps in two ways. Users see where they are in the hierarchy and can go up a level with one click. Google also gets another structured signal about the relationships between pages. You can also connect breadcrumbs to structured data to increase the chance of rich snippets in the SERP. In most CMSs breadcrumbs can be switched on with a single toggle. There is no reason not to have them. With some custom solutions it can sometimes be a battle with developers. Unfortunately.

URL structure should reflect the website’s hierarchy. A clean URL: web.cz/kategorie/podkategorie/stranka. Not: web.cz/p?id=4827&cat=12. The URL is a signal for both Google and users. If you can tell from the URL where you are on the website, the structure works. Use short, descriptive URLs with a keyword. No unnecessary parameters, session IDs or hashes. I also do not recommend including the category in product URLs. In my view, a better solution is to use something like /p/ or /e/ in the URLs of all products, so it is clear at first glance that the page is a product. Also, do not leave a 404 page at .cz/e. Ideally, put the complete e-shop with all products there (but again, without creating duplicates).

Internal linking is probably the most underrated part of site structure. Menus and breadcrumbs are automatic. But contextual links in the text, such as “read also the article about X”, are what connect thematically related content and distribute link juice where you need it. In audits, this is usually the weakest point. Websites have 500 articles and almost no links between them in the text. Yet these are the very links that give Google a strong signal that the content is related and that the website has depth on the topic. A good SEO copywriter handles this automatically while writing, linking each article to 2 to 10 related ones. It is not just about SEO. You write about a particular topic, and it makes sense to connect it with the rest so that the pages can be browsed systematically.

A sitemap comes in two forms. An XML sitemap is for Googlebot: a list of all the URLs you want indexed. An HTML sitemap is for users: a clear page showing the structure of the website. An XML sitemap is a must. An HTML sitemap is useful on larger websites where the navigation cannot cover everything. Tip: the XML sitemap should contain only the URLs you want in the index. No noindex pages, 301 redirects or 404s. Ideally, it should update automatically.

How to design a website structure step by step

Here is how I do it in practice. The whole process has become much faster since Claude Code arrived. Six steps that work both for a new website and for a redesign of an existing one. It is nothing groundbreaking, but this systematic approach is exactly what separates a website with a well-thought-out architecture from one where pages were created chaotically without a plan.

1. Start with keyword research, not design. Most people start by sketching what the website should look like. That is wrong. The structure should be based on what people search for (keywords and phrases) and also on where you want to be. So I start with Ahrefs, Marketing Miner and Keywords Explorer and enter the client’s main topics. Keyword research tells you which categories make sense, which subcategories to create and what content belongs under them. Without it, you are designing the structure blind. It often happens that a client wants a category with zero search volume, while ignoring a topic that people search for a lot. More than other SEO consultants, I look into the business (what earns you the most now and what you want to earn you the most). I look at categories not only in terms of search volume, but I also try to find figures on how much money flows through each category and at what margin. That is crucial. If the client runs Google Ads heavily, that also often shows where the centre of gravity of the segment lies.

2. Clustering. After collecting keywords from up to dozens of sources, I export everything to Google Sheets. Then I group them into topic clusters, either manually or with Claude Code. Claude Code is excellent at this: I give it a list of 500 keywords and it sorts them into logical groups faster than I could with scripts. Each cluster = a potential category or page. This is where you see whether you have enough content for a separate category or whether a subpage is enough.

3. Draw the tree (or generate it). Open Google Sheets and create a simple hierarchy: column A = main categories, column B = subcategories, column C = individual pages. Nothing complicated, but you will see the structure visually. Some branches will be too deep, others too flat, and this is where you start balancing. The goal is for no branch to have more than 3 levels below the homepage and for important pages not to be hidden at the end of a long path. This is, of course, the e-commerce scenario, where everything is clearer and simpler. With B2B, you have more freedom, because competitors often get it wrong and you will not easily find inspiration anywhere.

4. Apply the 3-click rule. Every important page should be reachable within 3 clicks of the homepage. Not 5, not 7. Three. If you have a page that visitors only reach after 4+ clicks, either move it higher in the hierarchy or add direct links to it from other pages. This rule is not an absolute dogma, but it is a good benchmark. Core things should, of course, be reachable straight away. When I land on a website, the layout and hierarchy should tell me where the business makes its money and roughly in what percentages. In reality, of course, few people think this way.

5. Plan your internal linking. A tree is not enough. You also need “shortcuts”: contextual links between pages that are thematically related but are not in the same branch of the tree. This is the pillar-cluster model: the pillar page (category) links to the cluster pages (articles), and they link back to the pillar.

6. Validate with data. Once your structure is designed (or rebuilt), validate it. Screaming Frog runs a crawl and shows you crawl depth (how many clicks each page is from the homepage). I filter pages with a depth of 4+ and consider whether they really need to be that deep. In Ahrefs Site Audit, you see the same plus orphan pages (pages without a single internal link). Screaming Frog also shows internal linking (how many links point to each page and from where). A simple rule of thumb applies here: the six most important pages for your business should have the most internal links, and structural hub pages should, of course, rank high too.

The most common mistakes I see in site structures

Over years of auditing, some problems recur so regularly that I could give them names. Here are the most common ones, and most of them are relatively easy to fix once you know where to look. One crawl in Screaming Frog is enough to see how many of them your website has.

The website structure does not reflect the business: Your business priorities should largely be mirrored on the website. Then add peripheral topics related to your main ones so that you fully cover your own segment.

Poor naming of services or categories: Very often, clients have two services instead of 6, or they do not break their services down further. The same goes for e-commerce categories. It pays to go more granular and cover less frequent queries.

Orphan pages: pages with no internal links pointing to them. They exist and are in the XML sitemap, but in terms of internal structure they are invisible. Google can find them through the sitemap, but without internal links it gives them no authority. Larger websites often have hundreds of them. Screaming Frog finds them in a few minutes; just compare the crawl with the XML sitemap.

Hierarchy that is too deep: 5, 6, sometimes even 7 levels of nesting. I have seen an e-shop where you reach a product via: homepage → section → category → subcategory → sub-subcategory → filter → product.

Overloaded navigation: 40+ menu items. The idea is good (I want to show everything), the result is bad (the visitor sees nothing). With more than 7 or 8 menu items, attention gets lost. Less is more.

Duplicate paths: the same content available at different URLs. Category A → page X and also category B → page X. Google does not know which version to index and often indexes the wrong one. Or none at all. The solution is a canonical tag pointing to the preferred version, but the best solution, of course, is to have one version of each page and redirect any duplicates.

Tags without logic: many CMSs allow tagging, but nobody plans the taxonomy in advance. The result: 200 tags, 150 of which have a single article. Pointless pages that cannibalise the main categories and dilute the crawl budget.

No contextual links: websites where internal linking happens only through menus and breadcrumbs. No links to related content in the body of articles. This means you lose the strongest tool for distributing PageRank, as well as the chance to keep visitors on the website longer.

Poorly targeted link building and PR: This is not strictly about structure, but when I look at which pages clients target with link building or online PR, there is a mismatch. They often push the blog instead of the homepage and service pages. In my view, it is also a mistake when link building is too concentrated and everything points to 1 or 2 pages.

Structure for an e-shop vs. a company website

Each type of website has different needs and different limits. You cannot take an e-shop structure and transfer it to a company website, or vice versa. Here is how I think about it.

An e-shop needs the most robust structure. Typically: homepage → main categories (e.g. Clothing, Footwear, Accessories) → subcategories (Men’s shoes, Women’s shoes) → parametric filters (Size, Colour, Brand, Type) → product detail page. Tags can then cover a different angle, such as occasions to use the products for. With filters, the key decision is whether to let them be indexed. If filter pages target long-tail keywords with search volume, index them. If not, noindex. In my experience, it is better to let even combinations be indexed that show no visible search volume, as long as the combination makes sense. On the other hand, it is important not to create tens of thousands of pages with zero potential. Correctly indexed filters can bring a large volume of long-tail traffic. On e-shops with thousands of products, structure is literally make-or-break for organic traffic.

A company website is typically simpler. Homepage → Services → Individual services → Case studies / References. Plus About, Contact and Blog. Two levels are usually enough, and the whole site has fewer than 50 pages. A mistake I see often: company websites describe their services in one paragraph on a single page, instead of giving each service its own landing page with dedicated content. Some large companies, on the other hand, describe all their services on one page and end up targeting almost nothing. One paragraph won’t rank for anything. Every service you want to bring organic traffic to needs its own page with unique content, a title tag, internal links and a clear idea, and it has to be competitive. For company websites, it is also important that the structure reflects the branding: the brand’s main message and direction should be clear from the navigation and page hierarchy alone.

—

The key is not to think of structure as a one-off decision. The website grows, content is added and the offering changes. The structure has to evolve with it. Once every six months, open Screaming Frog, run a crawl and check whether any branch has grown in a way that no longer makes sense or no longer matches your business priorities. Or whether you have 50 orphan pages you don’t know about.

In the same way, I track in Ahrefs how organic traffic changes across individual sections of the website. If a whole category is stagnating, the problem is often not the content but the structure. Within digital marketing, website structure is one of the cheapest ways to move the numbers. What’s more, structure now affects everything else too. For GEO and visibility in LLMs, a more robust structure is demonstrably better. The same has been true for years for paid campaigns, which then send traffic to pages within the website structure.

Website structure is not a project with an end. It is ongoing work, like building a house. But if you get it right at the start, you will save yourself hours of painful restructuring later. A good structure is the foundation everything else stands on: content, link building, conversions, customers, numbers. Without it, you can only grow up to a certain point. With it, everything scales better.