AgentSkillsCN

visual-testing

通过 5 秒测试、偏好测试、可取性测试以及行为分析等方法,全面评估视觉设计的有效性。

SKILL.md
--- frontmatter
title: Visual Design Testing
description: Methods for testing visual design effectiveness including 5-second tests, preference testing, desirability testing, and behavioral analysis
tags:
  - frontend
  - design-testing
  - ux-research
  - visual-design
  - staylook
  - attitudinal-testing
  - behavioral-testing
name: visual-testing

Visual Design Testing Skill

Based on NN/g Methodology: Attitudinal and Behavioral Testing for Visual Design

What is it?

This skill provides structured methods for testing visual design effectiveness—from capturing first impressions with 5-second tests to measuring brand perception through desirability testing, and validating design decisions with A/B testing.

Why use it?

  • Validate Assumptions: You and your team are not the users—test with real audience
  • Brand Alignment: Verify design expresses intended brand traits
  • Reduce Risk: Catch perception issues before development
  • Data-Driven Decisions: Choose between design variations with evidence
  • Measure Impact: Understand how visual design affects user behavior

Part 1: When to Test Visual Design

Testing Throughout the Design Process

PhasePurposeMethodsFidelityParticipants
EarlyValidate direction5-second, preferenceLow-medium5-8/variation
MidRefine variationsPreference, word choiceMedium-high8-12/variation
Pre-LaunchFinal validationA/B testing, usabilityProduction12+ for behavioral

Part 2: Testing Methods Overview

Attitudinal Methods (Self-Reported)

Gather thoughts, feelings, and opinions to evaluate brand alignment.

5-Second Test

  • Purpose: Capture gut reaction, first impression
  • Duration: Show design exactly 5 seconds
  • Key Rule: Do NOT warn about 5-second limit
  • Questions After: What do you remember? What stood out? How did it feel?
  • Sample Size: 5-15 per design

First-Click Test

  • Purpose: Test if users find what they need
  • Method: Show static image, ask "where would you click to..."
  • Measure: Success rate, click heat maps
  • Sample Size: 8-12 per design

Preference Test

  • Purpose: Compare 2-3 design variations
  • Key Rule: Differences must be obvious to non-designers
  • Key Rule: Randomize order shown to participants
  • Questions: Which do you prefer? Why? Which feels more [trait]?
  • Sample Size: 8-15 total

Desirability Testing (Word Choice)

  • Purpose: Verify specific brand traits are perceived
  • Method: Closed word list (20-30 words)
  • Categories: Target traits, opposite traits, distractors
  • Key Rule: Allow viewing design while choosing words
  • Sample Size: 8-20 per design

Rating Scales

  • Purpose: Quantitative comparison data
  • Format: 5-point semantic differential (e.g., Cheap ←→ Premium)
  • Analysis: Mean, median, standard deviation
  • Sample Size: 15+ for statistical analysis

Behavioral Methods (Observed)

Observe how users interact with the design.

Eyetracking

  • Purpose: Track where users look
  • Reveals: Fixations, gaze paths, heat maps
  • Alternative: First-click testing, recall questions

A/B Testing

  • Purpose: Measure real behavioral impact
  • Metrics: CTR, conversion, time on page, bounce rate
  • Key Rule: Change one major element at a time
  • Sample Size: 100s-1000s (use calculators)

Part 3: Staylook-Specific Testing

Testing the "One Highlight" Principle

QuestionTesting Method
Is THE Expressive element noticed first?5-second: "What stood out?"
Does it draw action/clicks?First-click test
Is there confusion from competing elements?Preference test: 1 highlight vs multiple

Testing Radius Hierarchy

QuestionTesting Method
Does nesting create visual grouping?Open word choice for "organized", "structured"
Is curved aesthetic perceived as warm?Rating scale: Sharp ←→ Curved

Testing Standard vs Expressive Balance

QuestionTesting Method
Is 90/10 ratio balanced?Desirability testing
Is Expressive special and attention-grabbing?5-second test, first-click

Part 4: Method Selection Guide

GoalBest Method
Test first impression5-Second Test
Test findability/hierarchyFirst-Click Test
Compare design optionsPreference Testing
Verify brand traits perceivedClosed Word Choice
Discover unknown perceptionsOpen Word Choice
Get quantitative dataRating Scales
Measure real behaviorA/B Testing
Understand attention flowEyetracking

Part 5: Sample Sizes Quick Reference

MethodMinimumRecommendedNotes
5-Second Test510-15Per design variation
First-Click Test510-15Per design variation
Preference Test812-15Total (see all variations)
Open Word Choice815-20Per design variation
Closed Word Choice815-20Per design variation
Rating Scales1530+For statistical analysis
A/B Testing100s-1000sVariesUse sample size calculator

Part 6: Common Pitfalls

PitfallWhy It's a Problem
Warning about 5-second testsPrimes unnatural memorization
Showing similar designsDifferences must be obvious to non-designers
Order biasAlways randomize/counterbalance design order
Leading questionsDon't ask "Do you like this?"
Small samples for quantitativeGet 15+ for rating scales
Aesthetic before usabilityAlways test behavior first in combined studies
Too many words in closed choiceKeep list to 20-30 words

Part 7: Best Practices

Do:

  • Test with real target users, not colleagues
  • Use structured methods, not just "do you like it?"
  • Randomize order of design variations
  • Allow design viewing during word selection
  • Combine multiple methods for robust insights
  • Test early and iterate

Don't:

  • Assume your taste matches users
  • Test subtle differences imperceptible to non-designers
  • Ask about aesthetics before usability testing
  • Use only one method
  • Skip follow-up "why" questions
  • Test once and consider it validated

Further Resources


Visual Design Testing Skill — Based on NN/g Methodology