Confirmation dialog

Problem Context

• The user has already invoked an action and the product must verify intent before committing it.
• The consequence is destructive, costly, privacy-affecting, security-sensitive, externally visible, or difficult to reverse.
• A simple undo path cannot reliably restore the prior state after the action completes.
• The dialog can add information that might reasonably cause the user to cancel.

Selection Rules

• Choose confirmation when the action has a clear reason not to proceed and the user may change their mind after seeing the consequence.
• Prefer undo when the exact prior state can be restored reliably after a frequent low-risk action.
• Prefer prevention when the action is dangerous mainly because it is placed too close to routine controls.
• Use a typed confirmation or equivalent extra step for broad account, workspace, production-data, or permission deletion.
• Do not confirm ordinary save, archive, dismiss, filter, navigation, or low-risk edits that users perform repeatedly.
• Do not use confirmation copy to teach users about a feature; use inline education, preview, or feedback instead.
• Use a regular modal task when the user needs to edit values, not merely decide whether the consequence should happen.

Required States

• Pre-action state with an explicit consequential trigger.
• Decision state that names the affected object, scope, and irreversible or costly consequence.
• Safe cancellation state that leaves the object unchanged and returns focus to the trigger.
• Blocked destructive state before required typed text or safety condition is satisfied.
• Confirmed committing state after the user deliberately activates the destructive action.
• Outcome state that reports deletion, cancellation, or policy-blocked status.
• Repeated-action state that avoids prompting for routine reversible actions.

Interaction Contract

• The confirmation appears only after a user-initiated action, not as surprise interruption unrelated to the current command.
• The dialog has alertdialog semantics or equivalent platform behavior when it communicates an urgent decision that requires a response.
• The title, description, and final action label name what will happen, not just ask Are you sure.
• The safe path is always reachable by keyboard and pointer before the destructive path is committed.
• Escape cancels when cancellation is safe and follows the same unchanged-object path as the cancel button.
• The destructive action is disabled or unavailable until any required typed confirmation, checkbox, or acknowledgement is satisfied.
• Closing after cancel or completion returns focus to the trigger or to the next logical outcome location.
• The dialog is not repeated for high-frequency actions where users will learn to dismiss it automatically.

Implementation Checklist

• Name the affected object, count, or scope in the title or main instruction.
• Explain the consequence once, including whether recovery, audit, billing, permissions, or external recipients are affected.
• Use a safe cancel label that explains the safe outcome, such as Keep archive or Cancel deletion.
• Use a destructive action label that contains the verb and object, such as Delete archive.
• Avoid OK, Yes, No, and Cancel-only button pairs when they force users to reinterpret the question.
• Keep the safe action reachable by default and avoid committing destruction from an accidental Enter key.
• Require typed text or a deliberate extra step for severe or broad-scope deletion.
• Return focus predictably after cancel, Escape, and completion.
• Track whether frequent prompts are being dismissed unread and replace them with undo or prevention when possible.

Common Generated-UI Mistakes

• Asking users to confirm every routine action until they stop reading.
• Using vague Are you sure copy without object, scope, or consequence.
• Making the destructive action the easiest accidental target.
• Using confirmation instead of undo for fast reversible actions.
• Using undo language for irreversible actions that cannot truly be restored.
• Showing an alert dialog that only educates, advertises, or repeats what the trigger already said.

Critique Questions

• What information in this dialog could reasonably make a user cancel?
• Can the action be safely undone instead of blocked beforehand?
• Does the destructive button say exactly what will happen?
• Is the safe path reachable without interpreting a double negative?
• Is the action frequent enough that this prompt will become automatic?
• Does the dialog name the object, count, permission, cost, recipient, or external system affected?