Architecture
Macro topic, clusters, pages, micro-intents and internal links.
A semantic site build, from the first intent to measurement.
A site is designed as a system: the business goal, the topical map, the URL structure, the content, the code, the images, the structured data and the analytics all have to say the same thing.
What is included?
Content
Direct answers, sources, tables, stated limits and calls to action.
Implementation
Semantic HTML, accessibility, responsive CSS and a cohesive JSON-LD graph.
Launch and measurement
Sitemap, robots, validation, a Search Console and analytics plan, and a baseline period.
Preparation for AI search
Pages are written so that the answer stands as a self-contained unit, because a passage is what enters an AI answer, not a document.
What does a build cost?
from €1,200priced by map size
The figure follows the number of pages in the topical map and whether the site is monolingual or bilingual. The map is drawn and agreed before the first page is written, so the price rests on a known scope.
What is included
- Topical map: macro topic, clusters, pages, micro-intents and internal links.
- Content written with direct answers, sources and explicitly labelled limits.
- Semantic HTML, accessibility, responsive CSS and a cohesive JSON-LD graph.
- Hand-written diagrams in the site palette, with descriptions for screen readers.
- Sitemap, robots, validation, Search Console and GA4 setup, and a defined baseline period.
What is not included
- Hosting, domain and photography licences.
- Maintenance and new content after launch.
- Campaigns, ads and link building.
What is not promised: architecture is a condition, not a guarantee. The outcome depends on demand in the niche, on competitors, and on what happens to the site after launch.
Which projects is it for?
It makes most sense for local businesses, aggregators, specialist services and e-commerce projects that hold verifiable data and want an architecture that stays maintainable.
It makes least sense where the content does not yet exist and no data supports it. A well-built structure around empty pages is still empty pages.
The delivery flow is described in detail on the process page.
The same principle applies to AI answers: a passage is selected rather than a document, so every page has to carry a self-contained answer. What that means in practice is set out on the page about AI search and SEO.
What is decided before any code is written
- The measurable business function. What has to happen on the site for it to be worth building.
- The evidence you hold. Local data, prices, tests and experience — the things that cannot be paraphrased from elsewhere.
- The demand threshold that decides whether a query gets its own URL.
- The terminology. One agreed name per concept, before the first page is written.
- The baseline period against which anything will later be measured.
Skipping these produces a site that has to be restructured after the first month of data, which is more expensive than deciding them at the start.
| Proposal | What it fails on | Where the content went |
|---|---|---|
| A page per city-to-city route | Demand — a handful of searches a year per route | A row in the intercity fare table |
| A page per taxi company | Low distinctness — the companies share the same attributes | A card in the list on the city hub |
| A separate page for the night tariff | Low distinctness — same entity, same predicate | A row in the fare table |
| A page per city district | Demand and entity — a district has no fare of its own | Not covered anywhere |
The first row is the difference between a site of a few hundred pages and a site of a few thousand empty ones. The full decision log, including one call that did not work out, is in the study of how the map was built.
When does a build not make sense?
A build assumes there is a business goal, a source for the data the content carries, and someone maintaining the site after launch. Without those you get a site that starts going stale on day one.
- The existing site works and only lacks structure. An audit is cheaper and faster — findings at URL level show what changes without rewriting everything.
- There is no source for the data. Content that carries numbers needs somewhere those numbers are refreshed from. Without it, the page is accurate only on the day it ships.
- The site is needed within a week. A map and its content take time; a fast site without a map is work that gets redone in six months.
- A result is expected without maintenance. A site with no new content and no measurement after launch holds its position; it does not grow.
In those four cases we say so at the first conversation.
Send your project for review
Describe the project and you will get a proposed next step, not a generic package.