A taxonomy should leave room for future content, not fill up around what is known today

What it is

If a classification exactly covers the objects that exist today, the first new product tomorrow has no cell. Leaving room means cutting on an extensible dimension—leaving values that have not appeared yet but must appear on the same dimension—rather than tiling a jigsaw around the current inventory. Filling the present writes “what we have now” as “what we are allowed to have later.” Leaving room is not minting empty categories in the nav to look prepared. It is a cut that can absorb new objects without an immediate tree change.

Empty nav items hurt scent. Room is left in the dimension and the rule; what appears in the UI should still be classes that have content.

Why it happens

Once a dimension is enumerated from present instances (one class per three products), a fourth product can only go to Other or force a tree change. Enumerated by attribute (product line, life-cycle state), a new instance is a new value or combination and the skeleton still holds. What is reserved is the extension of the rule, not empty boxes drawn today. A fill-the-present scheme looks complete in a content-planning meeting because it maps one-to-one onto the current roadmap; when the roadmap moves, the mapping voids, and migration cost is experienced as “sudden need to extend.”

People learn category rules. If the rule equals the current list, the learning will not generalize to new objects, and every launch reteaches the nav.

Studying it

Test filing generalization on unseen new objects, not only card sorts on the present corpus.

• Paradigms: build a tree on the current set, then introduce objects that were absent from the tasks (new region, new file type) and see whether they file under the original rule; contrast instance-named cuts with attribute-named cuts.
• Independent variables: instance names versus attribute names, presence of Other, whether the dimension is declared open.
• Dependent variables: share of new objects fileable without a tree change, forced new top-level classes, whether people can say where a new object “ought” to go.
• Methodological note: workshops that only discuss known content systematically pick fill-the-present schemes. Stress-test with objects that “could exist in six months.” Empty classes in a tree test get clicked and mislead; adding empties is not how room is reserved.

Where it stops holding

Closed enumerations (seven days, fixed tax codes) should be filled; leaving room is the error. A one-off campaign site has no “future content”; reserving for an imagined season two leaves dead classes. Leaving room is also not premature splitting into tiny classes—that is a different failure. This leaf only requires that the cut can still receive new values at the same grain.

Applying it

• Write the value domain of each layer: what attribute it cuts on, and whether that attribute is open or closed. Open layers must not use current instances as class names.
• Walk three “not yet existent but legal” objects through filing. If they cannot walk, change the cut; do not draw empty nav first.
• The nav shows currently non-empty classes; the rule layer keeps the ability to receive new values.
• Verify with the next item on the product roadmap, asking someone uninvolved where the current rule would put it. “Nowhere” or “we’d need a new top level” means the taxonomy was filled around today, not reserved along a dimension.

Related

• Within the group: G1.11.2 Premature classes for tiny sets lose meaning as content grows · G1.11.3 The cost of changing architecture rises with inbound links, bookmarks, and settled mental models · G1.11.4 Structural changes need redirects so old paths do not break · G1.11.5 Governance decides who may add or merge classes, or the taxonomy inflates without order · G1.11.6 Imbalanced class use is a signal to restructure; do not rely on intuition alone
• Adjacent: G1.03 Taxonomy and hierarchy · G1.02 Organization systems · G1.09 Types of organizational structures
• Search terms: extensible taxonomy · open facet · taxonomy growth