Finalist Presentation
GIF
Product Overview

🥇First Place Google x Rcfintech x Ut projects

Summary: Revo is an end to end return management system that lets companies customizable return eligibility and resale actions to fit business needs

REVO: Return Management Revolution

Inspiration

E-commerce returns represent a $166 million loss for every $1 billion in sales, with 5 billion pounds ending up in landfills annually. 43% of Gen Z consumers engage in "wardrobing" (buying, using once, then returning), and smaller apparel retailers like Toronto's Kotn and Peace Collective were especially vulnerable. These businesses lack enterprise-level resources yet face devastating return fraud impacts. We set out to transform this challenge into an opportunity.

The existing process is a labour intensive system that is prone to errors. Validating return policies and updating records is a manual process and items that could be resold or recycled are often mishandled at warehouses, leading to waste and missed profits.

What it does

REVO (Return, Verify, Optimize) is an AI-powered platform that:

Detects fraudulent returns through computer vision and pattern analysis
Creates customizable refund policies (full refund, store credit, exchanges)
Automatically routes returns to optimal channels (restock, resale, donation)
Provides actionable analytics on return trends and customer behaviours
Integrates seamlessly with e-commerce platforms like Shopify

How we built it

We built REVO with a modular architecture featuring:

Next.js frontend for an intuitive dashboard experience
Python backend with FastAPI for high-performance processing
Firebase database for real-time data synchronization
Google Gemini-2.0-Flash AI model for multimodal return analysis
Custom rule engine for flexible return policy implementation
Integration hooks for popular e-commerce platforms

Challenges we ran into

Our biggest challenges included:

Balancing fraud detection accuracy against false positives that harm legitimate customers
Developing computer vision models that could detect subtle signs of wear in garments
Creating a scalable solution accessible to smaller retailers with limited technical resources
Building a no-code rule engine that remained powerful yet intuitive
Maintaining user privacy while collecting sufficient data for fraud detection

Accomplishments that we're proud of

We're particularly proud of:

Achieving 76% reduction in wardrobing-related losses in our pilot tests
Developing a tiered fraud scoring system that dramatically reduces false positives
Creating an intuitive dashboard that delivers actionable insights without requiring data science expertise
Building a solution that offers 280% ROI within 12 months for small-to-medium retailers
Designing a system that reduces return-related carbon emissions by 15%

What we learned

Throughout this project, we gained valuable insights into:

The complex psychology behind different return behaviours (bracketing vs. wardrobing)
How return policies significantly impact customer lifetime value
The technical challenges of building computer vision models for textile analysis
The specific needs of smaller retailers versus enterprise operations
Balancing operational efficiency with excellent customer experience

What's next for REVO

Our roadmap includes:

Expanding AI capabilities to detect additional fraud patterns
Developing dedicated solutions for specialty retail categories (formal wear, luxury goods)
Creating a community knowledge base where retailers can share anonymized fraud patterns
Building advanced predictive analytics for inventory planning based on return data
Exploring partnerships with sustainable recycling operations for non-resellable items