A flat structure cuts depth and grows the sibling set

What it is

Flat navigation takes destinations that used to sit several clicks down and lays them out as peers on one level. Depth falls; the sibling count rises. A four-level tree with four children per node, pressed into one level, becomes dozens of side-by-side entries. The claim is about a change of shape—fewer descents, a longer row of neighbors—not about whether that row can be scanned, and not about when flattening is wise.

Typical forms: a top bar that lists every primary destination at once, a settings page that lays every group in one long list, an app with four or five peer tabs. The shared property: the current level is almost the destination level.

Why it happens

A hierarchy hides options inside folds, so only a handful of children are in view at once. Flattening opens the folds; the view is now the union of several former levels. Users skip the “choose again” pages, but they must discriminate inside a larger set. Structurally this converts sequential decisions into a set decision: nodes once reached by a path are now reached from a list.

Once the set grows, visual search starts from a different question. People stop asking “which child do I drill?” and ask “which of these is my destination?” Spatial memory shifts from “second from the left on level three” to “which slot on this strip.” More slots slow the first find; once the slot is learned, later visits can go there directly, skipping intermediate levels. Both the gain and the cost of a flat structure come from the same unfolding.

Studying it

Hold the targets constant and compare deep drill-down with a single level that lists every leaf. Log the change in depth and the change in sibling count separately; do not only report “flat is better.”

• Paradigms: controlled menu studies in the Miller/Kiger tradition, treating “flat” as a structural condition rather than an optimum; a flattened tree test in which every leaf appears at level one.
• Independent variables: depth, number of options at the first level (they covary; orthogonalize or report the conversion).
• Dependent variables: clicks to a leaf, time to first location, time on a later visit after the layout is learned.
• Methodological note: pressing a four-level tree into one level also changes label grain—intermediate class names vanish and leaf names step forward. If labels are not controlled, the study may be measuring naming, not structure.

Where it stops holding

When the option set swings with permissions, region, or season, the slots of a flat structure will not sit still; the row changes daily and spatial memory cannot form. Huge collections (encyclopedias, catalogues with tens of thousands of SKUs) cannot actually flatten to one level; the so-called flat version is usually a wider top with a tree still underneath. Voice and command palettes are not spatial unfoldings, so this depth-versus-sibling swap does not apply.

Applying it

• When you flatten, write two inventories: which intermediate pages disappear, and how many peers will share the first screen. Missing either is not yet a structural decision.
• Count items in a top bar or tab bar as “all primary destinations.” Items hidden in an overflow are not still flat.
• Verify by asking someone uninvolved to list, from the first screen only, every destination reachable without opening a submenu. The count should match the leaves you claim sit on the flat level. If many destinations still live one level down, the structure did not flatten; the first level only looks wider.

Related

• Within the group: G2.02.2 Flat navigation fits limited, parallel content · G2.02.3 Flattening moves choice cost from depth to breadth
• Adjacent: G2.04 Depth versus breadth tradeoff · E5.15 Number of navigation items · G1.09 Types of organization systems
• Search terms: flat navigation · one-level menu · breadth versus depth