Virtual Try-On in Fashion E-Commerce: An AR Success Story

Problem Definition

An eyewear e-commerce brand faced two core issues in its online channel. First, customers hesitated to buy because they did not know how glasses would look on their face. Second, a significant portion of buyers returned products citing "did not suit my face."

In physical stores, customers tried on glasses before deciding, but the online channel lacked this equivalent. The company tried recommendation algorithms based on face shapes, but the impact was limited.

Solution: AR Virtual Try-On

The company decided to integrate a web-based AR virtual try-on feature. Without needing an app download, the customer's camera activates directly on the product page and the selected eyewear model is virtually placed on the face.

Technical Implementation

• High-quality 3D models were created for each eyewear model
• Face recognition and tracking system positions glasses on the face in real time
• Color variant switching happens instantly on the 3D model
• Customers can take and share photos of their try-on

Challenges Encountered

1. Varying lighting conditions: Light in the customer's environment affected appearance. Solved with dynamic lighting adaptation.
2. Glasses size calibration: Proportion to face must be correct. Automatic sizing via face measurement algorithms.
3. Device compatibility: Performance issues on older devices. Static preview offered on unsupported devices.

Results

Six months after launching the AR virtual try-on feature:

• A notable portion of product page visitors used the AR try-on feature
• Users who tried AR had significantly higher purchase rates than those who did not
• "Did not suit my face" returns decreased substantially
• Social media sharing of virtual try-on screenshots provided organic reach

Lessons Learned

• Virtual try-on creates a strong impact for "try before you buy" product categories
• Starting with WebAR allows quick results without app development
• The social sharing feature became an unexpected organic growth channel