Home Appliance Recall Detection System
Company: WIN Home Inspection
Timeline: 2020-2021
Executive Summary: Developed comprehensive appliance recall detection system that replaced expensive third-party vendors, saving $1.2M+ annually while improving accuracy from 60% to 94%.
Situation
WIN franchisees were paying approximately $10 per inspection to a third-party vendor for appliance recall information. The vendor had frequent inaccuracies (60% accuracy), misused customer data, and represented over $1M in annual costs across all franchisees.
Task
The challenge was to:
- • Build in-house recall detection system
- • Achieve higher accuracy than existing vendors
- • Integrate seamlessly into inspection workflow
Action
- • Analyzed CPSC's complex recall database and designed normalized schema
- • Developed multi-step fuzzy matching algorithm using RecordLinkage with Jaro-Winkler scoring
- • Implemented AWS Textract OCR for automatic data extraction from appliance photos
- • Built serverless matching system with exact, partial, and potential match categories
- • Created consumer-friendly PDF reports with clear recall information
Result
The system eliminated vendor dependencies while dramatically improving accuracy and user experience. It became a key differentiator in winning new contracts and saved franchisees over $1.2M annually.
Key Metrics
- • $1.2M+ annual franchisee savings
- • 94% accuracy rate vs. 60% with previous vendor
- • 100K+ appliances processed in first year
- • 3,200+ recalled appliances identified
- • 15 minutes time savings per report
- • 15% increase in contract conversion