Feedback Loops
CanonicalConfidence
Cognitive Load
Low
Evidence
production validated
Impact
feature
Ethical Guardrail
Never leave any action without acknowledgment. Never delay feedback on critical actions.
Design Intent
People need to know their actions had an effect -- immediately. Without clear feedback, users feel lost, uncertain, or ignored. Feedback Loops closes the loop between user action and system response.
Psychology Principle
People need to know their actions had an effect -- immediately. Without clear feedback, users feel lost.
Description
Close the loop between user action and system response with immediate, clear, and appropriate feedback.
When to use
Every user action -- taps, submissions, saves, deletes, navigation, agent actions.
Example
Figma Real-Time Collaboration: Cursor appears instantly for other users + live color-coded presence dots + instant visual updates with no perceptible delay.
Autonomy Compatibility
Behavioral Objective
Users feel in control and confident because every action receives immediate, clear confirmation.
- Reduced repeated actions and frustration
- Higher trust in the system
- Faster overall task completion
Target Actor
role
Everyday user
environment
Real-time interactions
emotional baseline
Needs reassurance that actions worked
ai familiarity
medium
risk tolerance
low
Execution Model
immediate_acknowledgment
Confirm the action was received instantly.
User clicks again because nothing happened.
result_communication
Show what actually changed.
User doesn't know if the action succeeded.
next_step_guidance
Tell the user what happens now or what to do next.
User is left hanging after success.
Failure Modes
Delayed or missing feedback
Enforce under 300ms response with skeleton states
Feedback is too subtle or technical
Use plain language and obvious visuals
Over-notification fatigue
Group or suppress secondary feedback
Negative feedback feels punishing
Keep tone helpful and calm
Feedback disappears too quickly
Allow users to review recent activity
Agent Decision Protocol
Triggers
- Users repeat clicks or actions
- Confusion spikes in a flow
- User asks 'Did that save?' or 'What happened?'
Escalation Strategy
L1: Diagnose missing or delayed feedback via behavioral_signals
L2: Nudge -- add instant toast or animation for the action
L3: Restructure -- redesign the feedback chain with acknowledgment + result + next step
L4: Constrain -- enforce performance budgets for sub-300ms response
L5: Yield -- flag for human UX designer review
Example
User taps Save -> instant 'Saved to your project!' toast + checkmark animation + updated timestamp.
Behavioral KPIs
Primary
- Repeated action rate (should approach zero)
- Task completion speed
- User confidence in system responsiveness
Risk
- Support tickets about 'nothing happened'
- Abandoned actions due to uncertainty
Trust
- User-reported 'I always know what's going on'
- Autonomy Dial usage when agent provides feedback
Behavioral Signals
missing_feedback
action_acknowledged=false AND user_repeated_action=true AND repeat_interval < 3s
action_completed=true AND feedback_shown=false
feedback_effective
action_acknowledged=true AND user_repeated_action=false
loop_closed=true AND next_action_taken=true AND dwell_time < 5s
feedback_fatigue
feedback_shown=true AND feedback_dismissed=true AND dismiss_rate > 50%
notification_count_per_session > 15 AND engagement_rate < 20%
Decay Monitoring
Revalidate when
- New interaction surfaces are added
- Performance budgets change
- User expectations for responsiveness increase
Decay signals
- Rising repeated-action rate
- User confusion about system state
- Feedback that 'the app feels laggy'
Pattern Relationships
Amplifies
Conflicts with