Back to lab notesBuild notes
Build notes3 min read

A name is part of the interface.

Labels should explain the object or action in front of someone. A short word is not always a clear one.

A button labelled “Clear” looks wonderfully economical. Four letters. Plenty of breathing room.

Clear what?

Suppose an imaginary seating-plan tool has selected guests, a table assignment, and a filter hiding some names. “Clear” could remove the selection, unseat the guests, or reset the filter. Each meaning is plausible. Only one matches the code behind the button.

The interface has left the person to supply the missing noun.

I would rather give a control a slightly longer label than ask someone to discover its meaning by using it. “Clear selection” and “Remove table assignment” are different actions. They deserve different names even if both make something disappear from the screen.

Naming a control early can expose a problem in the design. If no plain label fits, the control may be trying to do several jobs at once.

In the seating-plan example, a guest is not an invitation and an invitation is not a table assignment. One invitation might cover several guests. Removing an assignment should not delete the guest record. If all three are called an “entry,” the interface has hidden distinctions the product still needs to handle.

Start by naming the objects in a plain description of the task. Then use those names consistently where the objects appear. Technical storage terms can remain in the code; the person arranging a dinner does not need to learn what the database calls a row.

Let the label survive outside its favorite screen

A word may seem clear beside the surrounding layout and become vague in a dialog, a notification, or a narrow mobile view.

For the hypothetical planner, “Remove” beside a table row might look obvious. “Remove guest from Table 6” in the confirmation is more precise. It also leaves less room to confuse that action with removing the person from the event altogether.

A useful description can include the consequence: “The guest will remain on the guest list.” Only promise that if it matches the behavior. Words cannot repair a delete operation that removes more than the interface says.

Names should also agree across the route. If the first screen calls something a table assignment, help text should not switch to placement and the result message to allocation. Variation makes prose pleasant; it can make controls harder to follow.

There is still room for personality. A playful page title can welcome someone into the tool. A consequential button needs to explain its action. The two do not have to compete.

During a design review, collect the labels from one task and read them without the screenshots. Can you tell which object each action affects? Then look at the screens and check that a label means the same thing in every location.

Ask someone unfamiliar with the interface what they expect a button to do before they press it. A hesitation is useful information, not an invitation to explain the design more loudly.

For the seating planner, the useful outcome is simple: someone can move a guest without wondering whether they are cancelling the invitation. A few extra letters are a small price for keeping that distinction clear.