Site silo structure

What is wrong with a typical website

Open almost any site in a niche with a real product or service — manufacturing, wholesale, B2B services, e‑commerce, government suppliers. B2B, B2C or B2G — the picture is often the same: a catalog or service list, delivery and payment pages, an “about” section, certificates. At best, returns and warranty.

That is not a website. It is a digital business card.

One niche example — logic for any site. I use workwear as an example: the product is complex, buyers may be individuals, legal entities or public buyers, and the assortment easily turns into an endless meaningless catalog. The same principles for sections and pages mapped to real demand apply to e‑commerce, services, B2B, media, etc.: entities and URLs change, not the principles. CMS, language and hosting do not change the architectural rules.

From an SEO perspective, such a structure means the site effectively fights for a limited set of roughly 15–100 head terms (niche-dependent). Everything below that — long-tail queries with specific intent — is missed. Someone searching for “fire‑retardant winter welders’ workwear buy” will land where that query has its own page.

Owners often know the product deeply but forget that visitors may not yet be experts. The site should expose both the product or service and the business and people behind it: who you are, how production or service works, what backs your promises. Thoughtful topical architecture (silo) helps both search engines and users move from general to specific without gaps.

Semantic coverage in numbers

Semantic coverage: the metric you were missing

Before comparing schemes, here is a working definition that keeps the rest simple.

Semantic coverage is the share of real search demand on a topic that your site covers with its pages. Not page count, not rankings — intent coverage: how fully the structure answers what people actually search for.

It is not an official term — it is a practical metric for design work. Close cousins in the industry are semantic cocoons and intent-based cluster coverage; both overlap with how teams look at content gaps and topical completeness.

From definition to execution

Three URL schemes for one site: from chaos to silo

Below are three URL trees for one hypothetical project (using the workwear logic above): typical chaos, basic silo and extended silo. Compare them by how many real intents each version can cover, not by URL depth for aesthetics.

Scheme 1 — “As usual”. No silo, no structure

site.com/
├── catalog/
│   ├── product-1/
│   └── product-2/
├── delivery/          ← изолированная страница без родителя
├── payment/           ← то же самое
├── about/
└── contacts/

What is wrong:

/delivery/ and /payment/ sit at the second level but there is no /clients/ hub. No hierarchy, no breadcrumbs, no logic. The crawler sees pages, not structure.

The catalog is flat. Categories are not tied to services, materials or industries.

Semantic coverage — roughly 5–15% of real demand.

Nested routes matter

URL depth is not cosmetic. A crawler treats /catalog/spetsodezhda/zimnyaya/dlya-svarshchikov/ as part of a topical cluster. A flat /zimnyaya-spetsodezhda-dlya-svarshchikov/ is just a page.

Rule: every URL level must be a real page with its own content, not a technical segment.

Sitemap: an index file plus separate maps per section (static, catalog, services, blog, etc.) make it easier to control URL volume and changes — seohead.tech does the same.

Scheme 2 — Basic silo

site.com/
├── catalog/                          ← хаб каталога
│   ├── spetsodezhda/
│   │   ├── zimnyaya/
│   │   └── letnyaya/
│   └── siz/
│       ├── zashchita-golovy/
│       └── zashchita-ruk/
│
├── services/                         ← хаб услуг
│   ├── poshiv-pod-zakaz/
│   └── naneseniye-logotipa/
│
├── clients/                          ← хаб для клиента
│   ├── delivery/
│   ├── payment/
│   └── garantiya/
│
├── about/
│   ├── team/
│   └── sertifikaty/
│
└── blog/

What you gain:

A hub for each major area. /clients/ is the parent for delivery, payment and warranty. Users and crawlers see hierarchy; URLs match structure; breadcrumbs make sense.

Still missing:

• No pages for SEO filters (winter + fire‑retardant + welders)
• No industry landings
• No materials section, glossary or calculators
• Sections are fully isolated — no cross-links
• Author is unclear; weak E‑E‑A‑T

Semantic coverage — roughly 20–30%.

Scheme 3 — Extended silo. Full coverage

site.com/
│
├── catalog/
│   ├── spetsodezhda/
│   │   ├── zimnyaya/
│   │   ├── letnyaya/
│   │   └── dlya-svarshchikov/
│   ├── signalnaya/
│   │   ├── zimnyaya/
│   │   └── letnyaya/
│   └── siz/
│       ├── zashchita-golovy/
│       ├── zashchita-ruk/
│       └── gotovyye-komplekty/
│           ├── komplekt-elektrika/
│           ├── komplekt-svarshchika/
│           └── komplekt-stroitelya/
│
├── services/
│   ├── poshiv-pod-zakaz/
│   │   ├── malyye-serii/
│   │   └── krupnyye-tirazhi/
│   └── naneseniye-logotipa/
│       ├── vyshivka/
│       ├── dtf-pechat/
│       └── sravneniye-metodov/
│
├── industries/
│   ├── stroitelstvo/
│   ├── neftegaz/
│   ├── pishchevaya/
│   └── zhkkh/
│
├── materials/                        ← отдельный шаблон, не блог
│   ├── khlobok/
│   ├── smesovaya-tkan/
│   │   ├── ts-65-35/
│   │   └── ts-50-50/
│   └── fr-tkani/
│       ├── nomex/
│       ├── proban/
│       └── sravneniye/
│
├── clients/
│   ├── delivery/
│   ├── payment/
│   ├── tablitsy-razmerov/
│   │   ├── muzhskaya/
│   │   └── zhenskaya/
│   ├── ukhod/
│   │   ├── kak-stirat/
│   │   └── fr-odezhda/
│   └── normy-vydachi-siz/
│
├── about/
│   ├── team/
│   │   └── direktor/
│   ├── proizvodstvo/
│   ├── sertifikaty/
│   └── vakansii/
│
├── cases/
│   ├── komplektatsiya-stroitelnogo-obekta/
│   └── spetsodezhda-dlya-zhkkh/
│
├── glossary/
│   ├── fr-odezhda/
│   ├── antistatika/
│   └── gost-12-4-303/
│
├── calculators/
│   ├── komplekt-dlya-rabochego/
│   └── normy-vydachi/
│
├── blog/
│
└── prajslist/

Semantic coverage — roughly 70–90%.

With this structure the site covers the full journey: composition and properties for newcomers; catalog or listings with filters for specific demand; services showing how you deliver outcomes; a client hub for logistics and terms; industry and case pages showing where it already worked.

When people hesitate they need proof — production or process, documents, guarantees, a real contact point online or offline. These are not “company pages”; they answer objections before sales talks to them.

Glossary and blog capture the long tail — from definitions to narrow expert scenarios. They attract people not ready to buy yet and build brand trust.

How to build site structure: a step-by-step guide

Architecture is designed before the CMS editor opens or the first page ships. The sequence is the same for any niche.

AI prompt: generate site structure

Copy the text below into any AI chat or agent: it will ask follow‑ups and return a ready .md with a section tree.

Ты — SEO-архитектор и контент-стратег. Твоя задача — спроектировать структуру сайта по принципу extended silo с полным семантическим охватом.

Прежде чем строить структуру, задай мне 3–5 уточняющих вопроса:
- Что продаёт или делает бизнес (продукт, услуга, ниша)?
- Кто целевая аудитория (B2B, B2C, B2G, смешанная)?
- Есть ли уже структура или сайт — если да, попроси её прислать?
- Какие отрасли или сегменты клиентов есть?
- Какие дополнительные разделы важны: блог, словарь, калькуляторы, кейсы, материалы/технологии?

После получения ответов сформируй структуру по следующим правилам:
1. Каждый раздел — отдельный смысловой хаб с дочерними страницами.
2. Каталог или перечень услуг — через матрицу (категория × атрибут), только там, где есть реальный спрос.
3. Обязательно включи: автор/эксперт, хаб «О компании», хаб «Клиентам», кейсы, словарь или блог.
4. Не создавай страницы ради объёма — только там, где есть интент.

Выдай результат строго в формате Markdown:
- # — сайт (домен или название)
- ## — верхний уровень (каталог, услуги, о компании и т.д.)
- ### — подраздел или хаб
- #### — листовые страницы

Только заголовки разделов — без URL, без описаний, без пояснений внутри структуры.

What happens next: the AI asks questions → you answer or send an existing structure → you get a draft .md → import to XMind or similar → refine visually → hand off as a developer reference.

Example output file (neutral labels)

Below is a shortened .md fragment with neutral headings — no niche or products — only the nesting skeleton. A full structure can contain hundreds of pages.

# project.example

## Ассортимент

### Линейка «Стандарт»
#### Вариант для среды A
#### Вариант для среды B
#### Универсальные комплекты

### Линейка «Профи»
#### Сценарий X
#### Сценарий Y
#### Сценарий Z

### Сопутствующие позиции
#### Защита зоны 1
#### Защита зоны 2
#### Готовые наборы

## Услуги

### Изготовление на заказ
#### Малые партии
#### Крупные партии

### Брендирование
#### Метод A
#### Метод B

## База знаний о составе

### Материал группы 1
#### Подтип α
#### Подтип β

### Материал группы 2
#### Подтип γ

## Сценарии применения

### Отрасль 1
### Отрасль 2
### Отрасль 3

## Клиентам

### Размеры и подбор
### Эксплуатация и уход
### Нормы и регламенты

## О компании

### Команда
#### Руководство

### Производство / площадки
### Документы и допуски

## Блог
## Словарь
## Прайс-лист
## Контакты

Import via File → Import → Markdown in XMind to get a nested map, then tune priorities and cross-links.

zoom_in

SEO filter matrix instead of an endless catalog

Use categories as the stable backbone and cover detail demand with SEO filters and attribute combinations — without a flat, blown-up catalog. You need a matrix: category × attribute. Each filled cell = one SEO page with its own URL, H1, meta and content. Empty cell = no page, no thin intent. Same logic for narrow or wide assortments: meaningful intersections for real demand, not infinite catalog branches.

That yields 40–60 targeted pages instead of five categories. Each is a concrete listing for a concrete query. A dash is an honest signal: no demand, no thin page for structure’s sake.

Cross-links: silo is not a prison

A common mistake is to read silo as “sections never link to each other.” What should stay isolated is navigational patterns. Contextual cross-links between sections create semantic relatedness.

Silo and link pyramids work together. Silo defines topical clusters and which blocks sit next to each other; the pyramid moves equity inside a cluster (hub → leaves → back). Cross-cluster links stay deliberate and precise — as in the examples below — without diluting the topic. In practice this is a hybrid model. External write-ups (KeyGroup on linking and crawling; “semantic fortress” on demand coverage) sit on the same axis.

Examples of good cross-links:

• Material page /materials/fr-tkani/nomex/ → link to /catalog/dlya-svarshchikov/ogneupornaya/
• Product card → material page
• Case study → products, materials, services
• /services/poshiv-pod-zakaz/ → link to /materials/
• /industries/stroitelstvo/ → links to catalog, cases, services

Note. When copy is substantive and structure is sound, links appear naturally where readers need them. A Nomex fabric page naturally mentions where it is used; a case links to materials and services. If you must force links, the structure or the writing is wrong.

A case study as a landing page, not a gallery

Typical “cases” pages: photo grids and a line like “We delivered.” No per-case URL, no narrative, no links.

A proper case is a landing page:

zoom_in

If there are work stages, show them as a gallery in “before / during / after”. Rough finish → install → finished space. Readers scroll and grasp scale. That is not only trust — it is content that keeps people on the page and earns real internal links between sections.

Such a page ranks for queries like “outfitting a construction site with workwear”, passes equity to product and service URLs, and matches contractor-intent.

Author and E-E-A-T: a human behind the site

Search quality signals increasingly punish generic, faceless content: if readers cannot see who speaks and what experience backs it, you lose to resources with verifiable authors and cases. That holds for B2B, B2C and B2G.

In niches with a real product or complex service, people want to know who stands behind the offer — a director, expert or lead specialist as the voice of the content, on any CMS and site type.

• Their photo on materials, articles and service pages
• Author page with bio, experience, production photos
• `author` in schema.org across templates
• Personal commentary instead of anonymous filler

This is not only an SEO signal — in long sales cycles, trust in the author converts.

Full coverage and the semantic core

When structure already covers the whole topic — glossary, materials, services, industries, cases — adjacent concepts are answered by architecture itself. “LSI” tools matter less, but demand analysis stays mandatory: without it you cannot prioritize pages or see where traffic really is.

SEO studies and practitioners who work with topical clusters and cocoons often see faster cluster rankings on deeply structured sites than on flat sites.

The semantic core is then less about “finding topics” — the structure already did — and more about prioritization: which pages to ship first, where demand is higher and competition lower.

Pros and cons of extended silo

Universal sections: a checklist for any site

This is a baseline that fits most businesses; tune by niche, audience and demand.

This is a minimum baseline. Depending on the business, the structure grows or shrinks — but the logic stays: each section maps to an intent type and connects to others through meaningful links.