Company Overview
- US-based B2B industrial ecommerce leader
- $16B annual revenue
Tech Overview
- 100M+ SKUs
- 1,000+ Categories
- Web, iOS and Android customer-facing apps
Business & Technical Challenges
- “Unlimited” assortment of similar parts confuses customers
- Customers “see” and “describe” products differently in text search, causing lost revenue and poor CSAT
- Key product attributes may be missing or incorrect
- While consumer products have a few key attributes, Industrial products have scores (or even hundreds) making text search challenging
- Text search engines (like Solr or Elasticsearch) are not optimized for industrial products making them less efficient
- Built extensive image library of parts and generated “synthetic” data to cover broken parts, poor lighting, etc.
- Developed extensive data science models to understand product “type”, and attributes like color, size, # of turns on a screw, etc. to find high-relevancy results
- Designed bounding-box algorithms to correctly identify multiple parts in an image
- Launched a cloud-based microservice that accepts images from mobile apps and returns high-quality results in 500ms
- Built real-time tracking and performance monitoring capabilities
- Established firm as an AI leader in industry and increased customer satisfaction
- Significantly improved product search experience – a key revenue driver
- Drove 20%+ growth in app download, usage and monthly-active-user count
- 4% improvement in market share
- 10+% improvement in mobile app promotional spend
- NPS score improved by 7 points
- Improved sales-rep productivity
- Improved inventory tracking and logging, bar-code scanning, etc.
