Feedback Design Skill | Agent Skills

Feedback Design

Feedback is how software communicates with users. Good feedback creates anticipation, confirms actions, and guides recovery.

Evidence Tiers

[Research]   — Peer-reviewed studies, controlled experiments
[Expert]     — Nielsen Norman Group, recognized UX authorities
[Case Study] — Documented examples from major products
[Convention] — Industry practice, limited formal validation

Multiple tags = stronger evidence: [Research][Expert]
Mixed findings noted as: [Research — Mixed]

Response Time Thresholds

[Research][Expert] Jakob Nielsen, based on Miller (1968), established response time limits in Usability Engineering (1993):

| Threshold | User Perception | |-----------|-----------------| | 0.1 sec | Feels instantaneous — direct manipulation illusion | | 1.0 sec | Noticeable delay — user stays focused but notices wait | | 10 sec | Attention limit — user needs progress indicator or will leave |

These thresholds are based on human perceptual abilities and remain foundational in interaction design.

Source: Nielsen Norman Group - Response Times

Progress Indicators

[Research] Dopamine research (Schultz, Sapolsky) shows the brain releases dopamine during anticipation of reward, not after. Progress indicators work because they create anticipation.

Pattern: Progress Over Spinners

Weak feedback:

Loading...

Strong feedback:

Uploading photo 3 of 7...
████████░░░░░░░░ 47%

Progress creates anticipation. Spinners create uncertainty.

Skeleton Screens

[Research — Mixed Results] Skeleton screen research shows inconsistent findings:

Mejtoft et al. (2018) found skeleton screens scored higher on perceived speed
Viget's study (136 participants) found skeleton screens performed worse than spinners — users took longer and evaluated wait time more negatively

When skeletons may help:

Familiar interfaces where users know what to expect
Very short wait times
Slow, steady animation (not rapid motion)

When spinners may be better:

Novel interfaces
Longer wait times
Users unfamiliar with the layout

Source: Viget - A Bone to Pick with Skeleton Screens

Immediate Acknowledgment

[Expert] Nielsen Norman and UX practitioners recommend immediate feedback for every user action:

| Timing | Feedback Type | |--------|---------------| | 0-100ms | Visual state change (button press, hover) | | 100ms-1s | Loading indicator if not complete | | 1-10s | Progress indicator with status | | 10s+ | Explanation + option to cancel |

Success Confirmation

[Convention] Acknowledge completion without over-celebrating.

Patronizing:

🎉 Great job! You did it! Your file was uploaded successfully!

Respectful:

File uploaded. 2.4 MB

Users need confirmation, not praise. Objective acknowledgment respects user intelligence.

Error Messages

[Expert] Nielsen's Heuristic #9: "Help users recognize, diagnose, and recover from errors."

Error messages should answer three questions:

What happened?
Why?
What can I do now?

Useless:

Error: Something went wrong

Actionable:

Upload failed: File exceeds 10MB limit
[Compress image] [Choose different file]

[Research] Cognitive Load Theory (Sweller, 1988) supports this: vague error messages increase extraneous cognitive load, forcing users to diagnose problems instead of solving them.

Source: Nielsen Norman - Error Message Guidelines

Optimistic UI

[Convention] Update UI immediately, reconcile with server afterward.

// Optimistic: Update UI first
updateUI()         // Instant feedback
sendToServer()     // Background
handleFailure()    // Rollback if needed

// Pessimistic: Wait for server
await sendToServer()  // User waits
updateUI()

Use when:

Success rate is very high (>99%)
Action is reversible
Failure can be gracefully handled

Caution: No formal research validates optimistic UI. It's practitioner convention based on perceived performance benefits.

Sound and Haptic Feedback

[Research] Studies on haptic feedback show tactile sensations can increase engagement (Apple's Taptic Engine research). However, overuse causes habituation.

[Convention] Use sparingly:

Completion of significant actions
Destructive actions requiring attention
Errors that need immediate notice

Avoid for:

Every button tap
Routine navigation
Background updates

Anti-Patterns With Research

Carousels / Auto-Rotating Sliders

Status: Generally ineffective Evidence: [Research] — Multiple studies with consistent findings

What research shows:

Notre Dame study: 1% click-through rate; 84% of clicks on first slide only
Search Engine Land: 0.65% CTR across B2B sites
Adobe test: Removing slider entirely increased sales 23%
Eye tracking (NNg): Users often skip carousels, perceiving them as ads ("banner blindness")

Why they fail:

Auto-rotation moves content before users can read it
Users don't trust rotating content (ad-like)
Most users see only slide 1
Creates "choice paralysis" — nothing feels primary

If you must use a carousel:

Don't auto-rotate (or use 7+ second intervals)
Pause on hover/interaction
Replace dots with meaningful labels
Make first slide count (84% of engagement)
Consider if static content would work better

Better alternatives:

Static hero with clear hierarchy
Tabbed content (user-controlled)
Scrolling content sections

Sources:

Key Sources

Nielsen, J. (1993). Usability Engineering. Academic Press.
Miller, R.B. (1968). Response time in man-computer conversational transactions.
Sweller, J. (1988). Cognitive load during problem solving.
Nielsen Norman Group - Response Times
Mejtoft et al. (2018) - Skeleton Screens Study
Viget - Skeleton Screens Research

Agent Skills: Feedback Design

Install this agent skill to your local

Skill Files