Layering must not make users guess entry location
Aliases: predictable entry point · layered navigation · entry location
What it is
Layered presentation genuinely reduces complexity only when people can predict where more relevant capability is, how to enter it, and when it appears. A predictable entry point means that a function absent from the current surface remains reachable without random browsing through stable location, clear naming, hierarchy, contextual cues, or a search path. Containment is not the problem; an entry that cannot be inferred is.
Why it happens
People create search hypotheses from task goals: is this a global setting, object-related operation, more options, or help? Consistent grouping, location, and naming let them narrow scope step by step. If similar capability is sometimes on top, sometimes in context menu, and sometimes on another page, they must traverse every container or depend on memory. Guessing cost extends tasks, increases error, and makes an unfound capability psychologically equivalent to absent.
Studying it
Give people goals requiring secondary or advanced capability but no path. Observe their first destination, containers explored, abandonment point, and explanation. Measure entry-prediction accuracy, navigation depth, reversal, search use, and help requests. When comparing architectures, examine not only eventual success but whether search paths follow the hierarchy cues the system provides.
Where it stops holding
Predictability does not require one entry for everything. Object operations, global preferences, and emergency recovery can reasonably live in different locations. The problem is when differences lack an understandable classification rule. Search can compensate for some unknown paths, but cannot hide every layering problem: if people do not know which word to search for, it helps little.
Applying it
- Organize layers around understandable objects, tasks, and scope, keeping location, names, and expansion behavior stable for the same category.
- Use summaries, more-actions labels, breadcrumbs, related entries, and contextual cues to make the next layer's existence and direction inferable.
- Conduct unguided navigation testing for advanced and exception tasks; when people cycle among wrong containers, repair the category, naming, or entry relationship.