Site architecture for service businesses

Why architecture decides what ranks

Most ranking conversations start with keywords and content. Those matter, but they sit on top of something quieter: the shape of the site. Which pages exist, what each is responsible for, how they are named, and how they point at one another. That shape is what a crawler experiences first and what a visitor experiences without noticing.

A service business usually grows a site by accretion. A page for the first service, then one for a second, then a landing page for a campaign, then a blog, then three pages for a new city. Nothing is wrong with any single page. The problem is that nobody decided how they relate, so two pages end up competing for the same query, the best page is reachable only from a footer link, and the crawler spends its visit on tag archives.

Architecture fixes that once. A clear structure tells search engines which page is the canonical answer for a job, tells answer engines which page to cite, and tells a visitor where they are. It also makes every future page cheaper to add because the slot already exists.

The three layers every service site needs

A service business has a small number of things it sells and, often, a number of places it sells them. The structure that holds up over years separates those concerns into three layers.

1. Service hubs. One page per service you want to be known for. This is the page you would hand a prospect who asked “what do you actually do?” It carries the full explanation, the proof, the process and the call to action. It is the canonical target for the broad query.
2. Service-plus-place pages. One page per service in a location where the service, the team or the logistics genuinely differ. A roofing company with crews in two counties can justify two pages. The same company with one crew and twelve city names cannot.
3. Reference pages. Guides, checklists, glossaries and explainers that answer the questions people have before they are ready to buy. They earn links and citations, and they link down into the hubs. This library is an example of the layer.

The layers link in a specific direction. Reference pages link to hubs. Hubs link to their place pages and to the reference material that supports them. Place pages link back up to their hub and sideways to the one or two sibling locations a visitor might actually be in. Nothing links to everything.

URL and folder rules that age well

URLs are the one part of a structure that is painful to change later, because every change costs a redirect and some of the link equity that was attached to the old address. Decide the rules before the pages exist.

• One folder per service, named for the service in plain words: /water-damage-restoration/ rather than /services/wdr-2/.
• Place pages live inside the service folder: /water-damage-restoration/round-rock/. The folder tells both crawler and reader what the page is a child of.
• Reference content lives in its own folder (/resources/ here) and is never nested under a service, even when it is about one. That keeps the service folder for pages that sell.
• Lowercase, hyphens, no dates, no parameters in canonical URLs. Google’s URL structure guidance says the same, and visitors read hyphens as spaces.
• Trailing slash or no trailing slash is a coin flip; pick one and redirect the other site-wide.

A single flat site, where every page sits at the root, is tempting because it is simple. It also throws away the signal a folder gives. Folders group related pages for crawlers, make analytics filterable by service, and let you set rules (templates, schema, breadcrumbs) per branch.

Internal links: the ranking lever you already own

External links are earned slowly and never fully controlled. Internal links are typed by you, cost nothing, and tell a crawler which pages you think matter most. Most service sites get this backwards: the privacy policy is linked from every page, and the highest-margin service page is linked from the navigation and nowhere else.

Three habits change that.

1. Link from body copy, not only from menus. A link inside a paragraph, with descriptive anchor text, carries context a menu link does not. Every reference page should link to at least one hub from within the text, where it is useful to the reader.
2. Reduce footer sprawl. Fifty footer links on every page dilute all fifty. Keep the footer to the pages a visitor needs from anywhere: contact, main services, legal.
3. Count the inbound links to each money page. If the service you most want to sell has fewer internal links than your blog tag archive, the site is telling Google the opposite of what you intend.

Location pages without the doorway problem

Google’s spam policies describe doorway pages as pages created to rank for specific, similar queries that funnel users to the same destination. A set of city pages that differ only in the place name is the textbook case, and the risk is not only a penalty: pages that say nothing new rarely rank well in the first place.

A location page is legitimate when it carries information that is only true in that place. Ask what a visitor in that town needs to know that the hub page cannot tell them.

• Who serves the area, with the actual team or office if there is one.
• Service differences: permit requirements, response times, local pricing factors, the kinds of jobs that area generates.
• Proof from that area: projects, reviews that mention the town, local affiliations.
• Practical logistics: coverage boundaries, how to book, what happens next.

If you cannot fill those sections honestly for a town, do not build the page. Serve the area from the hub with a service-area list and let the Google Business Profile carry the local signal. Fewer, fuller pages beat a wide spread of thin ones.

Pagination, filters and other crawl traps

Any index page that can be sorted or filtered can generate far more URLs than you have content. A resource library with six topics, five formats, three levels and four sort orders can produce hundreds of combinations from eighteen entries. Decide which of those combinations deserve to be indexed, and keep the rest out of the index without blocking the content they lead to.

• Index the clean topic views (/resources/topic/seo/) because they are useful landing pages with a stable set of entries.
• Let search, sort and page parameters exist for visitors, but point their canonical to the clean view, or mark them noindex, so they do not compete.
• Make sure every entry is linked from at least one indexable page. If an item is only reachable via page three of a filtered list, it is effectively orphaned.
• Follow Google’s pagination guidance: give each page its own URL, link pages sequentially, and avoid loading items only on scroll without a linked fallback.

Crawl budget is rarely the limiting factor for a service site, but crawl focus is. A crawler that spends its visit on filter permutations learns less about the pages you care about.

How to tell if the structure is working

Architecture does not show up as a single metric. It shows up as the absence of certain problems. Three signals, checked quarterly, are enough.

1. One page per query. In Search Console, filter a core query and look at which pages receive impressions for it. If two or more pages trade places week to week, they are competing and one should be merged or pointed at the other.
2. Coverage without noise. The indexed page count should be close to the number of pages you intend to rank. A large gap in either direction (hundreds of parameter URLs, or key pages marked “crawled, currently not indexed”) is a structure problem before it is a content problem.
3. Path to enquiry. In analytics, look at the pages visited before a contact form submission. The expected pattern is reference page to hub to contact. If visitors reach contact from a place page they entered on, fine. If they bounce between two similar pages first, the two pages are not distinct enough.

The thirty-minute architecture audit

Run this with the site open in one tab and a crawler export, or just the sitemap, in another. Tick what is true today. Anything unticked is the next piece of work.

0 of 12 ticked·Saved in this browser only

A site that ticks all twelve has an architecture that will carry hundreds more pages without a redesign. A site that ticks fewer than six is usually better served by restructuring before publishing anything new, because each new page inherits the confusion.

If you want a second pair of eyes on the result, the SEO service page describes how the work is scoped, and the web design service covers rebuilding a structure from the ground up.