Site architecture is the part of SEO that produces the largest gains for the least money and gets discussed the least, because it is unglamorous. Nobody sells an "internal linking package". There is no monthly report line for it. And yet on most sites we audit, redistributing internal links is the single change with the fastest measurable effect.
This article covers how search engines read a site's structure, the four architectural problems that account for most underperformance, and a concrete afternoon's work that reliably moves pages that were already close.
Every internal link does three things simultaneously, and understanding all three explains why the work matters.
It provides a discovery route. Crawlers follow links. A page reachable only through a sitemap is discovered slowly and revisited rarely.
It distributes authority. Whatever credibility your domain has flows along internal links. Pages that receive many internal links from important pages inherit more of it.
It declares importance and topic. This is the part people miss. The number of internal links a page receives, and the words used in them, are a direct statement from you about what that page is and how much it matters. If your most commercially important page receives two links from the footer, you have said it is a footer page.
Take the page you most want to rank. Count how many internal links point at it from pages that already receive traffic. If the answer is "the menu and nothing else", you have found the reason it underperforms, and the fix costs nothing.
| Problem | Symptom | Fix |
|---|---|---|
| Orphan pages | Published, indexed, no impressions, no internal links | Two or three contextual links from trafficked pages |
| Excessive depth | Important pages four or five clicks from home | Flatten: link directly from category or homepage |
| Homepage hoarding | The front page ranks for everything, deeper pages for nothing | Push authority outward with descriptive contextual links |
| Undifferentiated anchors | Every link says "read more" or the brand name | Describe the destination in the words a buyer would use |
Homepage hoarding deserves elaboration because it is so common on small local sites and so often misread as success. The front page accumulates almost all external links and all navigation links, so it ranks for service queries that should belong to service pages. It works in the short term and caps growth in the medium term, because one page cannot be genuinely relevant to fifteen different intents. The symptom is a site where the homepage gets 70% of organic traffic and every service page is stranded on page three.
You do not need a complex taxonomy. Three levels covers almost every case:
Brand, primary proposition, links to every service. Not a keyword landing page.
One page per service someone searches for separately. These are the commercial pages that must rank.
Question pages, guides, cases, district pages. Each links up to the service it supports.
The critical rule is that level 3 links up to level 2 with descriptive anchors, and level 2 links down to the level 3 pages that support it. That reciprocal cluster is what creates a recognisable topic on a small site — and it requires no external links at all, which is why it is available to every business immediately.
Avoid a fourth level unless the catalogue genuinely demands it. Every additional level halves the authority reaching the pages at the bottom and adds a click between the customer and the thing they want.
Anchor text is a signal about the destination, and the practical rules are short.
Describe what the reader will find, in the words a customer would use: "bathroom renovation costs in Valencia" rather than "click here" or "our services". Vary the phrasing naturally across links rather than repeating one exact phrase forty times — not because of a penalty threshold, but because identical repetition reads as automated to a person, and the pages that rank well read like they were written by one.
Two things to avoid. Linking the brand name to a service page tells search engines nothing about the page. And stuffing three links to the same destination into one paragraph adds nothing — the first one carries the signal.
This is a concrete exercise. It takes three to four hours on a typical local site and produces measurable movement within weeks.
Most architecture problems are not designed in. They accumulate. A site launches with a clean three-level structure, and two years later it has forty pages nobody planned, half of them competing with each other and a third reachable only through a search box.
Three habits prevent almost all of that drift, and they cost nothing to adopt.
Every new page gets its links at publication. Not "we will add internal links later", because later does not happen. The publishing checklist should include: which two existing pages link to this, with what anchor, and which page does this link up to. A page that cannot answer those three questions probably should not be published, because nothing in the structure will support it.
Check for cannibalisation before creating, not after. Before writing a new page, search your own site for the intent it targets. If an existing page already covers it, the correct action is usually to improve that page rather than to add a competitor to it. This single habit prevents the most expensive content mistake there is, and it takes two minutes.
Review the structure once a year. Not the content - the shape. Which pages have accumulated no links and no impressions in twelve months? Those are candidates for consolidation into a stronger page or removal. A smaller site where every page has a job outperforms a larger one where a third of the pages are dead weight, because the authority that was spread thinly gets concentrated.
Consolidation deserves a specific note, because businesses resist it. Merging three weak pages into one strong page and redirecting the other two feels like losing something. In practice it usually produces a page that outranks all three predecessors within a couple of months, because it finally has enough substance and enough links to compete. The instinct that more pages means more visibility is the single most reliable source of self-inflicted cannibalisation.
On a site of fifty pages this is irrelevant. On a site with filters, parameters, pagination or a large catalogue, structure decides where crawling goes — and crawling that goes to the wrong places is crawling that does not reach your new pages.
The observable version of this is the crawl log. If bots spend most of their visits on parameter URLs, tag archives and paginated listings while your service pages get visited every few days, the structure is leaking. The indexing log makes that visible directly rather than by inference, and it is the fastest way to confirm that a structural fix worked: crawl distribution shifts within days, long before rankings do.
The usual culprits on small sites are tag and category archives generated automatically by the CMS, author archives nobody reads, and date-based archives. None of them serve a search purpose on a local business site, and all of them absorb crawl and dilute internal links.
Two clarifications that save wasted effort.
The menu is not your architecture. It is a navigation convenience, and because it appears on every page, its links are treated as boilerplate and carry less weight than contextual ones. A page linked only from the menu is weakly linked, whatever the menu suggests visually. This is why trimming a menu during a redesign is dangerous but adding to it is not a fix.
Breadcrumbs help, modestly. They clarify hierarchy for both users and crawlers and can improve how your result is displayed. They are worth having and they are not a substitute for contextual linking.
Related-posts widgets sit somewhere in between: better than nothing, weaker than a link written into the sentence that gives the reader a reason to click.
Internal linking produces one of the clearest before-and-after signals in SEO, provided you recorded the date of the change.
Watch three things over the following six weeks. Impressions for the target pages should rise before positions do — the page becomes eligible for more queries as its topic becomes clearer. Position for the queries already close to the top ten should improve, because that is where a modest authority shift changes the outcome. And crawl frequency for the target pages should increase, visible in the indexing log within days.
What you should not expect: movement on head terms where you were on page four. Internal linking sharpens what is already close; it does not create relevance that was never there.
Pages with impressions but no internal links are the cheapest wins on any site. Your own analytics show them in ten minutes, free.
Sign in to Semalt See Fast IndexingThere is no magic number. Every important page should receive at least two or three contextual links from pages that already have traffic. Beyond that, link where it genuinely helps a reader.
Less than contextual ones. Site-wide boilerplate links are recognised as such. They are not useless, but they will not carry a page on their own.
Far less than externally, but identical repetition across forty links reads as automated. Vary it naturally and describe the destination.
On most local business sites, yes - or at least noindex them. They absorb crawl and dilute links without serving any search purpose.
Three clicks from the homepage is a safe target. Beyond that, expect it to be crawled less and treated as secondary unless something else compensates.
Internal linking is free, fast and almost entirely under your control, which makes it the best-value work available on most sites. It requires no outreach, no budget approval and no supplier — just an afternoon, a list of your commercially important pages and a list of your most-visited ones.
Do that comparison this week. Pull your top pages by clicks, set them against the pages you actually want to rank, and add the links that are missing. Record the date, and check impressions six weeks later.
If you would like the full architecture review — including the orphan list and the crawl distribution — get in touch. It is part of the free initial audit we deliver within 48 hours.
Free analytics for Google Search, SERP rankings and AI answers - plus fast indexing for your new pages.
Sign in to Semalt semalt.comGet in touch for a free audit
