SEO Taxonomy: How to Structure Categories, Tags and Facets

Most taxonomy problems are invisible until the site gets big enough to hurt.

A store launches with eight categories and forty products. Three years later it has 4,000 products, sixty categories, 900 tags, and filter combinations generating URLs faster than anyone can count them. Nobody decided that. It accumulated.

SEO taxonomy is the discipline of deciding it deliberately. This guide covers the structure types, the distinction between categories, tags, and facets that most sites get wrong, and the governance that stops it drifting again.

What SEO Taxonomy Actually Covers

Taxonomy is the classification system organising your content or catalogue into a navigable structure.

It is broader than categories. It includes tags, filters, attributes, and the internal linking that connects them. Categories are one component, not the whole thing.

Done well, it does three jobs at once: helps users find things, helps search engines understand relationships between pages, and keeps the site manageable as it grows. That third job is the one people undervalue until it is too late to fix cheaply.

The Four Structure Types

Hierarchical suits sites covering many related topics needing clear parent-child relationships. A recipe site with thousands of pages organises into broad groups then specific ones, because users need ways to narrow down.

Faceted suits catalogues where one item legitimately belongs under several attributes at once: region, type, colour, size, price. It is the most useful for users and the most dangerous for crawl budget, which we will come to.

The cluster model is the modern content approach. Rather than only categorising, pages connect through topical relationships and internal links, which helps search engines understand your site as an authority on a subject rather than a collection of separate pages.

Most real sites combine two or three of these.

Categories vs Tags vs Facets

This distinction is where most taxonomy goes wrong, so it is worth being precise.

Categories are the backbone. They are the gateways guiding visitors into more specific areas.

Tags link related items across categories. A T-shirt might be tagged cotton, graphic print, or crew neck. They enable discovery between things that live in different parts of the hierarchy.

Facets let users refine by multiple attributes at once without following a fixed path.

The mistake is treating these as interchangeable. Sites that use tags as extra categories end up with hundreds of near-empty archive pages competing with the real ones.

Depth and Breadth Rules

Two working conventions, both grounded in practical experience rather than any Google statement.

Start with no more than seven or eight top-level categories. More than that and navigation becomes unmanageable, particularly on mobile where nested dropdowns are genuinely painful.

Keep depth to three levels or fewer. Overcomplicated structures confuse users and bury pages. Where you have niche subcategories, consolidate them into broader parents and let filtering handle the refinement.

Categories should also be mutually exclusive at each level. If an item could sit equally in two sibling categories, your naming is ambiguous and users will feel it.

The depth rule and the click-depth rule reinforce each other. A page four levels deep in the taxonomy is usually four or five clicks from the homepage, which means it gets crawled rarely and receives little internal authority.

Which Taxonomy Pages Should Be Indexable

Not every page your taxonomy generates deserves to exist in the index.

The rule underneath: a taxonomy page earns indexation when people actually search for the thing it represents. “Waterproof jackets” is a query. “Waterproof jackets, size 12, blue, sorted by price ascending” is not.

For filtered category variations, canonical tags should point to the main category URL, so the signals consolidate rather than splitting across near-duplicates.

Mapping Keywords to Taxonomy Levels

Assign intent by level to prevent your own pages competing.

• Top-level categories target broad head terms
• Subcategories target mid-tail terms
• Product or article pages target long-tail specifics

So a category page targets “running shoes,” a subcategory targets “trail running shoes,” and a product page targets “women’s trail running shoes size 8.” Saving the specific phrases for product pages avoids cannibalisation between levels.

Check keyword difficulty alongside volume when choosing. A moderate-volume, low-difficulty term often delivers faster than a high-volume one you will chase for a year.

Category pages also need actual content. Reported analysis suggests category pages carrying 150 to 300 words of genuinely descriptive copy substantially outperform bare product grids, provided the copy addresses intent rather than listing product names.

The Tag Problem

Tags are the most abused element in most taxonomies.

They multiply silently, generate thin archive pages, and frequently duplicate category coverage. A blog with 900 tags has 900 pages of near-empty listings competing for attention with the content itself.

Sensible discipline:

• Three to five tags per item is usually enough
• Use tags for themes and attributes that cut across categories, not as substitute categories
• Noindex tag archives unless a specific tag has genuine search demand and enough items to be useful
• Audit periodically and merge or delete tags used fewer than a handful of times

The test: does this tag help someone discover related content they would not otherwise find? If not, it is metadata, not a page.

Faceted Navigation

The most common crawl budget problem on large sites.

Combining size, colour, and price generates enormous numbers of near-duplicate URLs, potentially millions on a large catalogue. Left crawlable, Googlebot spends its budget on filter permutations instead of your actual pages.

Manage it deliberately:

• Identify which single facets have real search demand and make those indexable with clean URLs
• Canonical multi-facet combinations to the parent category
• Block sort, view, and session parameters from crawl entirely
• Confirm assumptions with log file analysis rather than guessing

The goal is not to eliminate facets. Users need them. The goal is to decide which ones become pages.

Design for the Catalogue You Will Have

The most expensive taxonomy mistake is building for today’s size.

Adding categories later forces URL changes, which forces redirects, which risks rankings. Every restructure is a migration with all the associated hazards.

So plan for scale at the outset. Think about where the next 5,000 SKUs or 500 articles will sit before you need to place them. Build a structure with room in it rather than one that fits exactly.

Also design for mobile from the start. Deep nested dropdowns that work fine on desktop are unusable on a phone, and most traffic arrives on one.

Governance Keeps It From Drifting

This is the part almost nobody implements, and it is why taxonomies decay.

Document:

• Who can create a new category or tag
• Who approves attribute changes
• Naming conventions, including capitalisation and singular versus plural
• What qualifies something for its own category rather than a tag
• How to handle deprecation and redirects

Without rules, every team member adds what seems sensible to them, and three years later you have four tags meaning the same thing with slightly different spellings.

How Often to Review

At least twice a year, and immediately after any significant catalogue or content expansion.

A review should cover: empty or near-empty categories, tags used fewer than a handful of times, categories that have grown large enough to warrant subdivision, indexation status of taxonomy pages, and whether current naming still matches how people search.

Search behaviour shifts. A category named in 2022 may use terminology nobody types now.

Mistakes That Compound Over Time

• Treating tags as extra categories.
• Building for the current catalogue size rather than the future one.
• Leaving every facet combination crawlable.
• Nesting four or five levels deep because the hierarchy felt logical.
• Publishing category pages with no descriptive content.
• Targeting the same keywords at category and product level.
• Never writing down who is allowed to add a category.

Where to Start

Export a list of every category and tag on your site with the number of items in each.

Anything with fewer than three or four items is a candidate for merging or removal. Anything with hundreds is a candidate for subdivision.

That single spreadsheet usually reveals more structural problems than a full crawl, and it takes twenty minutes to produce.

FAQs