Why Categories Should Organise Products, Not Describe Them
Category trees become fragile when they are used as the place to store product facts. A retailer starts with Jackets, then adds Waterproof Jackets, Men's Waterproof Jackets, and Blue Men's Waterproof Jackets. Each branch answers a question that a structured product record should answer directly.
Navigation is a different concern
Categories help shoppers choose a starting point. They should be stable enough to support menus, landing pages and merchandising. They are not a durable encoding for colour, capacity, fit, material, compatibility or condition. Those are product attributes.
When attributes carry those facts, a jacket can sit in a sensible browse category while being found through any valid combination of waterproof rating, colour and size. The product is not copied. The category does not need a new branch for every commercial variation.
Keep the tree thin
A thin category tree is not an incomplete tree. It is a tree that refuses to absorb information owned elsewhere. The architectural rule is simple: categories organise navigation; templates define which product facts are required; governed attributes make those facts queryable.
A practical decision test
Before adding a category, ask whether it represents a stable browsing destination or a product fact. Lighting may be a useful destination. Brass table lamps is usually a combination of product type, finish and placement. The first belongs in navigation; the latter should be resolved by structured attributes, a collection, or a search result.
This distinction keeps change local. A merchandising team can create a seasonal landing page without changing product contracts. An engineering team can introduce a new attribute without rebuilding menus. The catalog gains a clear source of truth instead of a hierarchy that has to explain every possible product combination.
Stable categories, flexible experiences
Thin categories do not limit merchandising. They create a dependable foundation for it. Collections, landing pages, campaigns and search rules can assemble products for a particular audience without permanently encoding that audience into the taxonomy.
Consider a retailer promoting “small-space dining.” The selection may span folding tables, benches and compact chairs. Creating that phrase as a structural parent category would imply those products are fundamentally defined by the campaign. A collection can instead use governed dimensions, seating capacity and product type to assemble the range. When the campaign ends, no product has to be reclassified.
Separate stable navigation from changing merchandising
Thin category → Governed product attributes → Dynamic collection → Campaign experience
The operational cost of a deep tree
Every category branch creates maintenance. Products need assignment rules. Redirects and breadcrumbs need managing. Analytics inherit a new reporting dimension. Integrations must decide whether the path is meaningful or merely presentational. When attributes are encoded in the hierarchy, a correction to one product fact can trigger reclassification across several branches.
Deep trees also conceal gaps. A retailer may believe it supports waterproof filtering because it has a Waterproof Jackets category, even though waterproofing is not represented consistently on the products. The navigation provides the appearance of structure while preventing other channels from reusing the fact.
Governance questions for a category review
An effective taxonomy review should ask:
- Would a shopper reasonably begin a journey at this category?
- Is the concept stable across seasons and channels?
- Does the branch describe a product type, or merely an attribute value?
- Could the same result be generated from governed attributes?
- Are products duplicated or multiply assigned to compensate for discovery gaps?
- Do downstream teams treat the path as a product contract?
The goal is not the smallest possible tree. It is a tree in which every branch has a clear navigational purpose. Product meaning then remains available outside the tree—to search, filters, store systems, APIs and future channels.
The architecture in one rule
Categories answer “where can I browse?” Templates answer “what kind of product is this?” Attributes answer “what is true about it?” Keeping those questions separate gives retailers a stable taxonomy without sacrificing rich discovery.