Vague labels force people to probe items one by one

What it is

If labels cannot eliminate any neighbour (“Solutions,” “Services,” “Platform” in a row), people cannot choose before clicking and can only open them one by one. Probing postpones elimination that should have happened from the preview until after arrival. Each probe costs a load, an attention reset, and a return. Vague is not “unprofessional.” It is the absence of opposable previews among siblings, so choice degrades into traversal.

Accurate, distinctive labels let people skip most slots. Vague labels make every slot a required open. This is not “one wrong arrival hurts trust”: here there is not even an expectation that could be falsified, because the expectation is empty.

Why it happens

Elimination needs a difference. When several labels are all “possible” for the same target, working memory cannot rank them, and the strategy becomes left-to-right or top-to-bottom opens until the target is hit or the person quits. The wider the layer, the more expensive the probing; vague labels turn breadth from one scan into N entries. After returning, spatial memory still has not formed (each interior was inscrutable), so the next round is still a near-random order.

Search becomes an escape: not because the target was a better known-item search, but because browse’s preview function has stopped. It looks like “users prefer search.” The root is an unreadable menu.

Studying it

Count probes. Do not only score whether the target was eventually found.

• Paradigms: a layer of vague labels versus a layer of opposable labels, known-item tasks; record length of open-return chains, switches to search, whether people can afterwards state the difference among slots.
• Independent variables: eliminability of same-layer labels, layer width, presence of a short subtitle.
• Dependent variables: wrong slots opened before the target, total time, search switches, reports of “I could only click them one by one.”
• Methodological note: final success hides probe cost—people find it because they traversed. Count wrong opens as a primary metric. Long gaze trips among labels with no click is an empty preview that cannot even fire the first probe.

Where it stops holding

Experts who already choose by position, not by word, are not hurt on the first click by vague labels; new and occasional users still are. Imperative micro-labels (Save, Export, Delete) are already highly opposed; adding explanatory nouns makes them vaguer. Exploratory tasks are supposed to involve opening; probing is not always failure. Primary nav, though, is a predictable main path and should not lean on exploration. Spoken menus can read only one option at a time, so vagueness costs hearing the whole list—more expensive than visual traversal.

Applying it

• Run an elimination test on any two labels in a layer: hide the destination, ask which one you would rule out to find X. Pairs that cannot eliminate: change the words before talking about widening.
• Do not patch a vague slot with a hover explanation. The preview must hold unopened and unhovered.
• If more than half the labels in a layer cannot eliminate each other, the layer is not navigation. It is a drawer of things to open. Merge, or recut on a distinctive opposition.
• Verify by recording a known-item task and counting nav items opened and left before arrival. More than one, concentrated in one layer, is serial probing. Swap that layer for words that eliminate in opposite directions; probe count should drop. Changing only icons or color will not move the count.

Related

• Within the group: G2.08.1 A nav label must accurately preview what it opens · G2.08.2 A mismatch between expectation and arrival lowers trust in the whole site
• Adjacent: G1.04 Labeling systems · G2.02 Flat navigation · G2.04 Depth versus breadth tradeoff
• Search terms: vague labels · serial probing · information scent