Reverse a prior action, absorbing the anxiety of mistakes and encouraging confident exploration.
Structure
- Trigger
- Feedback
- Scope
Levels: single-action vs. multiple levels of undo
Whether undo can be applied only once or successively.
Linear vs. non-linear undo
-
Single-action undo: Reverses only the most recent action.
-
Multi-action undo: Allows reversing multiple actions sequentially.
-
Selective undo: Allows to pick a specific action from a history to undo, without affecting subsequent actions.
-
Regional undo: Enables undoing operations within specific regions or contexts
-
Branching models: Creates tree-like structures where actors can explore alternative paths
Collaboration
TODO:
To-do
Related pattern to capture: version history, a more durable counterpart to undo’s transient reversal, navigating named past states rather than stepping back through recent actions.
Resources & references
Related patterns
Enacts
- Temporality — lets the actor reverse an action after the fact, loosening commitment over time
Complements
- Transient feedback — the dismissal window doubles as the recovery window — transient feedback is the primary carrier for undo affordances
- Wizard — mitigates the high commitment level by allowing reversal of completed steps
Alternatives
- Inline confirmation — moves the safety net after the act instead of before it; prefer undo when the action reverses cleanly
Related
- Abort — related non-destructive exit; offer undo and restore affordances when state has been modified
- Action consequences — the strongest severity reducer — recovery after the act removes the need for confirmation before it
- Modality
- Saving — Reversibility in saved states
- Status feedback — provides immediate feedback on the undo action
- Command menu — can be used to access undo options
- Activity log — displays a history of actions that can be undone