Putting things off the platform convention raises search cost

What it is

On iOS, Back is expected top-left. A new app puts it top-right and parks Settings top-left. The skilled user’s first landing is top-left, and they are in Settings; a full search of the chrome then finds Back. The cost of violating a convention is not “it looks odd”. It is the extra miss, the extra mis-entry, the extra search of the whole shell.

Moving the hamburger from top-left to top-right, putting a dialog’s primary on the left, making close a × in the middle of the tab bar — all turn an already-automatic first landing into controlled search.

Why it happens

Search starts at the expected slot. The command is not in it, the program errors, and attention switches to item-by-item scanning. The scan set is the whole chrome — top bar, tab bar, edges — an order of magnitude larger than “look at one slot”. If the expected slot now holds another command, the bill includes a mis-execution: Settings opens, a draft is shared, a payment is tapped. Mis-execution is dearer than two extra seconds of search.

When a brand wants to look unlike the others, these slots are what get moved, because a slot is cheap to change and immediately visible. Cheap to change; expensive on every returning user’s first tap.

Studying it

Skilled users of that platform perform the same set of navigation commands in two layouts: matching the convention versus swapping or moving the key commands to the opposite side. No practice.

Independent variables: whether the command’s slot matches the platform, whether the expected slot holds another command, presence of a text label. Dependent variables: whether the first attempt hits, mis-executions, time to find the target, number of slots scanned (eye tracking).

A label can compensate in part, but the first landing is usually finished before the label is read, so “we wrote the word Back” does not stop that mis-entry.

Where it stops holding

A platform that does not yet have this convention (a brand-new device shell) cannot be violated; place has to be carried by other cues. A professional tool the user was trained on (a DAW’s idiosyncratic slots) may violate consumer conventions, but that is another community’s convention, not the absence of one. A one-shot marketing page has no muscle for Back; the cost of violation is mostly failing to find close — still expensive, the mechanism being search rather than mis-execution. Assistive tech walks in order, not from an expected slot; the cost of violation becomes extra focus steps, not a wrong tap.

Applying it

• Before changing chrome, list the slots that platform already owns (close, back, more, primary). A new product leaves them alone by default.
• Do not put another high-cost command in the expected slot. If top-left is not Back, leave it empty rather than parking Delete or Send there.
• When a violation is unavoidable, keep the moved command reachable in both the old and the new slot for a while, and make whatever now sits in the old slot reversible.
• How to check: people who only use that platform, this UI, immediately “leave this layer”. A first landing on the wrong command, or a tour of the chrome after the first glance, is the violation already charging.

Related

• Same group: F1.14.1 Positions acquire fixed meanings through long use, such as top-right for close or more · F1.14.3 Conventions follow platform and writing direction; they cannot be copied globally · F1.14.4 A new convention takes long, consistent exposure before it holds
• Nearby: F1.06 Positional stability of frequent actions · E1.18 Button order and platform convention
• Search terms: convention violation cost · unexpected slot · visual search