Skip to main content
Built-in A/B testing tools are coming soon. Currently, you’ll need to use your existing A/B testing platform (Google Optimize, VWO, Optimizely, etc.) to test virtual try-on variations. This guide shows you what to test and how to set up experiments.

​
Why A/B Test Virtual Try-On

A/B testing helps you optimize virtual try-on for maximum impact. Small changes can drive 10-30% improvements in engagement and conversions. How to A/B test now: Use your existing A/B testing tools (Google Optimize, Shopify’s built-in tests, or third-party apps) to test the variations recommended below.

​
What to A/B Test

​
What You Can Test

Button Copy

Test different call-to-action text via dashboard

Button Colors

Test different colors via dashboard

Product Selection

Test which products benefit most from try-on

Promotion Strategy

Test different ways to promote the feature

​
Other Tests

  • Product selection strategy (which products to disable)
  • Promotional tactics (homepage banners, emails, social media)
  • Multiple product image angles (max 4 per product)
  • Button corner position (top left vs top right)
Start with button copy tests – easy to do via dashboard and can improve engagement.

​
Test 1: Button Copy

​
Hypothesis

Question: What call-to-action text drives the highest engagement? Hypothesis: Action-oriented, benefit-focused copy (“See it on you”) will outperform generic copy (“Try On”).

​
Test Setup

Variant A (Control): “Try On” Variant B: “See it on you” Variant C: “Virtual try-on” Variant D: “Preview on yourself” Traffic split: 25% each Duration: 2 weeks

​
Success Metrics

Primary: Click-through rate on button Secondary:
  • Try-on completion rate
  • Conversion rate

​
Expected Results

Typical outcome:
  • Variant A: 35% engagement (baseline)
  • Variant B: 41% engagement (+17% – WINNER)
  • Variant C: 32% engagement (-9%)
  • Variant D: 37% engagement (+6%)
Winner: “See it on you” (more personal, benefit-focused)
Copy performance varies by audience. Test what resonates with YOUR customers.

​
Test 2: Supporting Messaging

​
Hypothesis

Question: Does adding context around the button increase engagement? Hypothesis: Brief explanatory text will increase engagement by reducing uncertainty.

​
Test Setup

Variant A (Control): Button only, no additional text Variant B: Button + text above: “See how this looks on you in seconds” Variant C: Button + text below: “AI-powered virtual try-on – no account needed” Variant D: Button + icon (camera) + text: 📷 “Try On virtually” Traffic split: 25% each Duration: 2 weeks

​
Success Metrics

Primary: Engagement rate Secondary:
  • Completion rate (did context set proper expectations?)
  • Conversion rate

​
Expected Results

Typical outcome:
  • Variant A: 35% engagement
  • Variant B: 41% engagement (+17% – WINNER)
  • Variant C: 37% engagement (+6%)
  • Variant D: 39% engagement (+11%)
Winner: Variant B (benefit-focused, addresses speed concern)

​
Test 4: Product Selection Strategy

​
Hypothesis

Question: Which products benefit most from virtual try-on? Hypothesis: Products with higher return rates will see bigger conversion lift from try-on.

​
Test Setup

Segment A: Products with < 15% return rate Segment B: Products with 15-25% return rate Segment C: Products with > 25% return rate Measure: Conversion rate lift for each segment Duration: 4 weeks (longer to account for returns)

​
Success Metrics

Primary: Conversion rate lift % Secondary:
  • Return rate reduction % (track via Shopify analytics, not Looksy dashboard)
  • ROI per segment

​
Expected Results

Typical outcome:
  • Segment A: 8% conversion lift, 10% return reduction
  • Segment B: 15% conversion lift, 20% return reduction
  • Segment C: 22% conversion lift, 35% return reduction (HIGHEST ROI)
Insight: Focus try-on on high-return products for maximum impact
Products with fit/style uncertainty benefit most from virtual try-on.

​
How to Run A/B Tests

​
Using Shopify Theme Editor

For button copy and styling:
  1. Create multiple product page templates with different button configurations
  2. Assign products randomly to each template
  3. Track performance via Looksy analytics
  4. Compare engagement rates

​
Using Third-Party Tools

Recommended tools:
  • Google Optimize (free, integrates with GA)
  • Optimizely (enterprise, advanced features)
  • VWO (visual editor, easy setup)
  • Convert (privacy-focused)
Setup:
  1. Install A/B testing tool on your store
  2. Create variants in the tool
  3. Set up goal tracking
  4. Run experiment

​
Native Looksy Testing (Pro Plan)

Built-in A/B testing features:
  • Button copy variants
  • Button styling tests
  • Automated winner selection
  • Statistical significance calculation
Access: Looksy Dashboard → Experiments
Use native Looksy testing if available – it’s pre-configured for virtual try-on optimization.

​
Statistical Significance

​
How Long to Run Tests

Minimum requirements:
  • At least 1,000 visitors per variant
  • At least 2 weeks (account for weekday/weekend patterns)
  • 95% statistical confidence before declaring a winner
Calculator: 
Required sample size =
  (Z-score² × p × (1-p)) / (margin of error²)

For 95% confidence, p=0.5, 5% margin:
  ≈ 385 visitors per variant minimum
Rule of thumb: Run tests until you have 1,000+ visitors per variant and clear winner emerges.
Declaring a winner too early leads to false positives. Be patient and wait for statistical significance.

​
Significance Calculators

Use online tools to verify significance:
  • Optimizely Stats Engine
  • VWO’s Bayesian calculator
  • AB Testguide calculator
Check: Is the p-value < 0.05? (95% confidence)

​
Analyzing Results

​
What to Look For

Clear winner:
  • One variant significantly outperforms others
  • Results are consistent across days
  • Statistical significance achieved
No clear winner:
  • Variants perform similarly
  • High variance in results
  • Need more data or different test
Unexpected results:
  • Variant performs worse than control
  • Inconsistent patterns
  • Check for implementation bugs

​
Making Decisions

If there’s a clear winner:
  1. Implement winning variant for all traffic
  2. Document learnings
  3. Plan next test
If results are inconclusive:
  1. Extend test duration
  2. Increase traffic to experiment
  3. Or move on to different test
If all variants underperform:
  1. Check for technical issues
  2. Review test setup
  3. Consider different hypothesis

​
Sequential Testing Strategy

Phase 1: Foundation (Weeks 1-4)
  1. Button placement
  2. Button copy
  3. Button styling
Phase 2: Optimization (Weeks 5-8) 4. Supporting messaging 5. Mobile-specific optimizations 6. Product selection strategy Phase 3: Advanced (Weeks 9-12) 7. Promotional tactics 8. Image quality improvements 9. Multi-variate tests (combine winners)
Test one element at a time. Don’t change multiple things simultaneously or you won’t know what drove the improvement.

​
Common A/B Testing Mistakes

​
Pitfalls to Avoid

Problem: Declaring winner before statistical significanceSolution: Wait for minimum sample size and 95% confidence
Problem: 6+ variants dilutes traffic, takes foreverSolution: Limit to 2-4 variants maximum
Problem: Adjusting variants during test invalidates resultsSolution: Plan thoroughly, don’t change once live
Problem: Traffic spike from campaign skews resultsSolution: Note external events, extend test if needed
Problem: Forgetting what was tested and whySolution: Maintain test log with hypotheses and results

​
Test Results Template

​
Document Every Test

Test Name: Button Placement - Above vs. Below Fold Date: Jan 15 - Jan 29, 2026 Hypothesis: Button above fold will increase engagement by 30% Variants:
  • Control: Button below description (28% engagement)
  • Variant B: Button above Add to Cart (38% engagement) ✓ WINNER
Results:
  • Sample size: 2,400 visitors per variant
  • Improvement: +35.7% engagement
  • Statistical significance: p < 0.01 (99% confidence)
  • Winner: Variant B
Action: Implement Variant B sitewide Learnings: Visibility matters more than placement within product details. Test mobile sticky bar next. Next test: Button copy optimization

​
Advanced: Multi-Variate Testing

​
Testing Multiple Elements Simultaneously

Example: Test button placement AND copy together Variants:
  • A: Above fold + “Try On”
  • B: Above fold + “See it on you”
  • C: Below fold + “Try On”
  • D: Below fold + “See it on you”
Pros: Find optimal combination faster Cons: Requires 4x more traffic Recommendation: Only do multi-variate after single-variable tests establish baselines

​
Next Steps

Dashboard Overview

Track your test results in real-time

Best Practices

Implement optimization strategies

Calculating ROI

Measure the impact of your tests

Custom Styling

Customize button styling for your brand