Most Sellers Guess. Top Sellers Test.
You spent hours creating product photos. But how do you know they're actually the BEST photos for conversions?
Most sellers pick images based on gut feeling. Top sellers A/B test systematically and let data decide. The difference in conversion rate can be 50-200%.
What to A/B Test
Main Image Variables
- Background: White vs light gray vs gradient
- Angle: Straight-on vs 45-degree vs top-down
- Product fill: 75% vs 85% vs 95% of frame
- Shadow: No shadow vs drop shadow vs contact shadow
- Styling: Plain product vs product with props
Secondary Image Variables
- Order: Which image comes second matters
- Lifestyle vs infographic: Which drives more engagement?
- Number of images: 5 vs 7 vs 9 images
- Video thumbnail: With or without video
A/B Testing on Amazon
Amazon's Manage Your Experiments
Available to Brand Registry sellers:
- Go to Seller Central → Brands → Manage Your Experiments
- Create a new experiment for your product
- Upload Version A (current) and Version B (new)
- Set duration (minimum 2 weeks recommended)
- Amazon splits traffic 50/50 and measures conversion rate
What Amazon Measures
- Conversion rate: % of visitors who purchase
- Units sold: Total units during the test
- Statistical significance: Whether the result is reliable
Amazon A/B Testing Tips
- Test ONE variable at a time
- Run for at least 4 weeks for reliable data
- Need ~1,000 sessions per variant for significance
- Don't change price or other variables during the test
A/B Testing on Shopify
Method 1: Shopify Apps
- Neat A/B Testing: Simple product image testing
- Intelligems: Advanced price and image testing
- Shoplift: Landing page and product page testing
Method 2: Manual Testing
- Run Image A for 2 weeks, record conversion rate
- Switch to Image B for 2 weeks, record conversion rate
- Compare results
Limitation: Not a true A/B test (time-based variables affect results). Use apps for accurate testing.
A/B Testing on eBay
eBay doesn't have built-in A/B testing, but you can:
- Create two identical listings with different images
- Run both for 2-4 weeks
- Compare views, watchers, and sales
- Keep the winner, end the loser
What the Data Shows
Based on aggregated A/B test results from e-commerce sellers:
Background
- White background wins for Amazon (required + highest CTR)
- Light gray beats white on Shopify by 5-8% (less harsh)
- Lifestyle backgrounds win on Etsy by 15-20%
Shadows
- Contact shadow beats no shadow by 8-12% on Amazon
- Drop shadow beats contact shadow for electronics by 5%
- Reflection shadow wins for luxury items by 10-15%
Product Fill
- 85% fill beats 70% fill by 12% on Amazon
- 80% fill beats 90% fill on Shopify by 7% (more breathing room)
Number of Images
- 7 images beats 3 images by 25% conversion rate
- 9 images beats 7 images by only 3% (diminishing returns)
- Adding video increases conversion by 15-20%
How to Create Test Variants Quickly
Using Pic1.ai, create multiple versions of the same product photo in minutes:
- Upload your product photo once
- Variant A: White background, contact shadow, 85% fill
- Variant B: White background, drop shadow, 85% fill
- Variant C: Light gray background, contact shadow, 80% fill
Each variant takes 30 seconds to create. Test all three and let data decide.
Statistical Significance: Don't Jump to Conclusions
Minimum Sample Size
- For a 5% conversion rate difference: ~1,500 visitors per variant
- For a 10% difference: ~400 visitors per variant
- For a 20% difference: ~100 visitors per variant
Common Mistakes
- Ending tests too early — Wait for statistical significance
- Testing during holidays — Seasonal traffic skews results
- Changing multiple variables — Test one thing at a time
- Ignoring mobile — 70%+ of e-commerce traffic is mobile
The Continuous Testing Framework
- Baseline: Measure current conversion rate
- Hypothesis: "A contact shadow will increase conversions by 10%"
- Test: Run A/B test for 4 weeks
- Analyze: Check statistical significance
- Implement: Use the winner
- Repeat: Test the next variable
Top sellers run continuous image tests. There's always something to optimize.
Create A/B test variants in seconds at pic1.ai/editor. Different backgrounds, shadows, and sizes from one upload.
