The SEO structure of a business website

The SEO structure of a business website has almost nothing to do with keywords. It has to do with how the information is divided up: how many pages exist, what each one contains, what address it sits at, and how they link to each other. Those decisions are taken in the first days of a project, before a single pixel of design exists, and they are the hardest to change afterwards.

A site with mediocre copy and good structure can rank. A site with excellent copy crammed into a single “Services” page cannot, however much you optimise. This article explains why, and what the alternative looks like — with a diagram of this site’s actual architecture, the decisions taken, and the reasons behind them.

What SEO structure means, and why it is not about keywords
The three-level architecture
How many pages a business website needs in order to appear in Google
One page, one search intent
Addresses: the rules, and what never changes
The heading hierarchy
Internal linking: sending authority to the pages that earn
Structured data
Sitemap, robots.txt and canonical
Category pages: the most commonly missed win
How a structure grows without breaking
The eight structural mistakes
Case study: the architecture of digital-art.ro

What SEO structure means, and why it is not about keywords

Google does not rank sites. It ranks pages. Every page competes separately, for its own question, against pages on other sites that answer the same question.

Everything that follows comes from that. If you have ten services described in ten paragraphs on a single page, you have one page trying to answer ten different questions. It will lose to ten pages, each dedicated to one — because each of those has its own title, its own address, and a whole piece of text on the subject.

What structure does, concretely

It decides what can be found. Information that has no page of its own cannot appear in results for its own search.
It distributes authority. Internal links carry the credibility a site has accumulated to the pages that matter.
It defines clarity. A site with a predictable hierarchy is easier to work through and to index.

Why it is irreversible

Structure becomes concrete in addresses. Changing one address means a redirect, reindexing and a period of instability. Changing a hundred addresses means a project in its own right, with a real risk of losing rankings. Which is why the structure is thought through once, at the start, and then left alone.

The three-level architecture

The model that works for almost any services company:

Level 1   /                         Home
Level 2   /services/                The umbrella page
Level 3   /services/<service>/      One page per service

Level 1 is the calling card. It does not compete on service terms — it competes on the company name, and on sending the visitor onwards quickly.

Level 2 is a real page, not a menu. It explains what you do as a whole and links to each service with a descriptive anchor. It can rank on general terms.

Level 3 is where the money is made. Every page has a title, an address and copy dedicated to a single service.

The three-click rule

No important page should be more than three clicks from the home page. Not because a robot would tire, but because depth reflects importance: what is buried at level five is treated as marginal.

What stays out of the hierarchy

The legal pages, the site map, the accessibility statement. They are necessary, they are linked from the footer, and they compete for nothing.

How many pages a business website needs in order to appear in Google

There is no minimum number. There is a rule: one page for each thing someone would search for separately. For a company with six services, that means around 12–15 pages: home, about, the services umbrella, six service pages, portfolio, FAQ, contact and the legal pages. Fewer pages does not mean a cleaner site; it means fewer ways in.

The number matters less than the distinction

Twenty pages saying roughly the same thing are worse than eight distinct ones. Near-duplicate content steals its own rankings — Google picks one and ignores the rest.

The test: if you cannot write 400 useful and different words for a new page, that page should not exist. Make it a section within an existing page.

One page, one search intent

The most broken rule in this whole article.

Search intent is what the person typing actually wants. The same words, different intents:

What they search	What they want	Which page answers
“business website design”	to find out what it involves and who does it	a service page
“how much does a business website cost”	a price range	an article
“business website examples”	to see work	a portfolio
“seo structure business website”	to learn how it is done	an article
“web design agency Constanța”	who to call	a local or contact page

Five intents, five pages. A single page trying to cover all of them answers none of them well.

How to check

Search the term in Google and look at what is already in the top five results. If they are service pages, the intent is commercial. If they are articles, it is informational. If it is a list with photographs, it is visual. Build the type of page that is already there — Google has already established what people expect.

Addresses: the rules, and what never changes

Rule	Yes	No
Real words, not identifiers	`/services/online-store/`	`/page?id=277`
No diacritics	`/services/identitate-vizuala/`	`/services/identitate-vizuală/`
Hyphens, not underscores	`/website-design/`	`/website_design/`
Short, without filler	`/about-us/`	`/the-about-us-and-team-page/`
The hierarchy visible in the address	`/services/packaging/`	`/packaging/`
Lower case, always	`/portfolio/`	`/Portfolio/`

The golden rule

A published address does not change. Not because it is ugly, not because you have found a better keyword, not because you have changed platform.

When we rebuilt this site, we kept 155 project addresses that look like /portfolio-items/<slug>/. That is an address inherited from a WordPress theme, not one we would have chosen. But they are 155 addresses that have been indexed for years, and renaming them would have meant 155 redirects for a gain no visitor would have noticed. They stayed exactly as they were.

When a change is unavoidable, you make a permanent 301 redirect from the old address to the new one, in a single step. Not in a chain.

The heading hierarchy

The headings in a page form a table of contents. It has to read like one.

H1   Business website design             ← one only, on every page
  H2 What a business website includes
    H3 The basic structure
    H3 The content
  H2 How we work
  H2 How long it takes

The three common mistakes

Several H1s. A table of contents does not have two main titles.
Skipped levels — an H1 followed straight by an H3. The most frequent defect of sites built with visual editors, which pick the level by how big the text looks.
Headings used as decoration. If a piece of text is a heading only so it will be bigger, it is not a heading; it is text at another size.

The practical rule: choose the level by position in the hierarchy, not by the size you want. Size comes from the styles.

An H2 phrased as a question

Searches have got longer — the average typed query has reached about 11 words, against 3–4 a few years ago (evergreen.media(opens in a new tab)). A heading phrased as a complete question, followed immediately by a 40–60 word answer, is the format engines extract most readily.

Internal linking: sending authority to the pages that earn

Internal links are the only ranking factor you control completely.

How it works

Every page has a quantity of credibility. Part of it passes on through each link. A page with many internal links pointing at it receives a clear signal: this matters.

Three rules

1. The anchor describes the destination. Not “click here” or “more”. The text in the link is a description of the target page: business website design.

2. Every article has one main target page. An article linking to seven service pages helps none of them. Choose one, link it with an exact anchor in the first third of the text, and add at most two or three links to related articles.

3. Links go both ways. When you publish a new article, add a link to it in older articles on the same subject. That is the step everyone forgets, and it is the one that turns isolated articles into a coherent cluster.

The topic cluster structure

        service page  ←── article 1
            ↑   ↑  ↑        ↕
     article 2  │  article 4 ←→ article 3

The articles link to each other and all of them point at the same commercial page. To a search engine, the cluster behaves like a single block of authority on a subject.

Structured data

Structured data is a summary written for machines, in JSON-LD, that says explicitly what is on the page. It does not change what the visitor sees. It changes how the result can look in search.

Type	Where it goes	What it brings
`Organization` / `ProfessionalService`	across the site	identifies the company as an entity: name, tax number, phone, address
`BreadcrumbList`	any page below the root	shows the path in the result, instead of the raw address
`Service`	each service page	connects the service to the provider
`FAQPage`	the FAQ page	can take up additional space in results
`BlogPosting`	each article	author, date, image
`WebSite`	the home page	the site’s name
`CreativeWork`	each portfolio project	describes the work

FAQPage has the best effort-to-result ratio of any of them: if you already have an FAQ page, marking it up is an hour’s work.

A warning: structured data has to describe what is actually on the page. Marking up questions that do not appear in the content breaks the rules and is penalised.

Sitemap, robots.txt and canonical

Three small files that often go wrong.

sitemap.xml — the complete list of addresses you want indexed, with the date each was last changed. It is generated automatically on every build; a manually maintained list becomes wrong within two months.

robots.txt — says what should not be crawled and points at the sitemap. The most frequent problem is inheritance: rules left over from an old platform that still block directories that no longer exist.

canonical — on every page, a tag declaring which is the official address. It resolves the situation where the same content is reachable at several addresses: with and without www, with and without a trailing /, with tracking parameters. Without a canonical, those compete with each other.

Category pages: the most commonly missed win

Any site with a collection — a portfolio, products, projects, services by field — needs real category pages, with addresses of their own.

What happens without them

You have a list of 150 projects and a filter applied in the page, through JavaScript. The address does not change when you filter. To a search engine there is one page: the complete list.

The result: you have nothing competing on category terms. You have pages for each individual project and one page for everything. Between them, nothing.

What changes with them

A category page:

Has its own address, so it can be indexed and ranked.
Can have its own introductory text about that type of work.
Becomes an intermediate node distributing authority: home page → category → project. Without it, the authority spreads evenly across 150 pages.
Reduces depth: every project comes within two clicks of the home page, not three or four.

How to build them properly

The filter stays, but as an enhancement, not as the main mechanism. The categories are real addresses that work without JavaScript; the in-page filter is a convenience for the visitor.

Every category gets 150–300 words of its own, not just a list. Without text, the page is a list of links and has nothing to compete with.

Do not create empty categories. A category with two items is a thin page competing with its own items. Below five or six entries, it does not deserve a page of its own.

How a structure grows without breaking

A site well structured at launch can degrade in two years. It happens through the accumulation of small decisions.

The rule for new pages

Before adding a page, three questions:

Is there a distinct search intent for it? If not, it is a section within an existing page.
Can I write 400 useful and different words? If not, the same conclusion.
Where does it attach in the hierarchy? A page with no link to it from the navigation is an orphan.

The signs that a structure is degrading

The menu has more than eight entries at the same level.
There are two pages that could answer the same search.
You have pages reachable only through internal search.
Adding a new page requires a decision about where to put it.
You have categories with a single entry.

The annual review

Once a year, list every address on the site alongside the impressions from Search Console. You will find three categories:

What you find	What to do
Pages with impressions and no clicks	rewrite the title and description
Pages with no impressions at all	check indexing; internal links may be missing
Pages cannibalising each other	merge them, with a 301 redirect from the weaker one

The last is the most valuable. Two mediocre pages on the same subject, merged into one good one, almost always rise.

The eight structural mistakes

A single “Services” page with everything you do, in paragraphs.
Duplicate pages created for keyword variations: “website design” and “website creation” as separate pages, with the same content. They cannibalise each other.
Filters that produce addresses with parameters, indexable, with near-identical content. Solved with a canonical.
The blog on another domain or a subdomain. Authority does not transfer the same way; the blog belongs at /blog/, on the same domain.
Excessive depth — pages five levels below the root.
Orphan pages, with no internal link pointing at them.
A menu built around the company’s org chart, not around how customers search. A customer searches for “installation”, not “the Technical Department”.
Addresses changed because somebody found a better keyword.

Case study: the architecture of digital-art.ro

The structure of this site, with the reasoning behind it:

/                                        Home
/despre-noi/                             About us
/servicii/                               The services umbrella
  /servicii/creare-site-de-prezentare/   ┐
  /servicii/magazin-online/              │
  /servicii/aplicatii-web-…/             │
  /servicii/inregistrare-domenii/        │ 10 service pages,
  /servicii/web-hosting/                 │ one for each
  /servicii/amazon-branding/             │ search intent
  /servicii/identitate-vizuala/          │
  /servicii/manual-de-identitate/        │
  /servicii/packaging/                   │
  /servicii/productie-publicitara/       ┘
/portofoliu/                             Index, 155 projects
  /portofoliu/webdesign-magazine-online/   94 projects
  /portofoliu/branding-identitate/         61 projects
/portfolio-items/<slug>/                 155 pages — addresses unchanged
/blog/                                   Article index
  /<article-slug>/                       the articles
/intrebari-frecvente/                    14 questions, marked up as FAQPage
/contact/

The four decisions that matter

1. Ten service pages, not one. Each has its own title, description and address. Ten chances of ranking instead of one.

2. Two category pages for the portfolio. There were 94 and 61 projects in two categories, with no page bringing them together. There are now two real, indexable addresses that can rank on category terms — and two additional nodes distributing authority to the projects.

3. The blog at /blog/, not at /category/blog/. The old address was a taxonomy archive used as a menu destination. It became a real page, with a 301 redirect from the old one.

4. The 155 project addresses were left untouched, even though we do not like them. They are the site’s most deeply indexed content. The first rule of structure is not to break what works.

Structure is settled at the start of a project and paid for once. Redone on an existing site, it costs several times as much and comes with the risk of losing rankings.

If you want to know where your site stands now, write to us — we will look over the architecture and tell you what can be fixed without a rebuild. See also what a business website project includes.