Global technology company
Multi-market competitive intelligence and fraud detection
Daily pipelines over millions of competitor data points, plus an entity resolution model improved market by market until the business trusted it.
- 10+ markets
- monitored daily in production
- Python, Spark, Entity resolution, Alerting & monitoring
- stack
Problem
- The client needed reliable daily competitive intelligence across more than ten markets to inform product, pricing, go-to-market, and in-platform fraud monitoring.
- Entity resolution of partners across heterogeneous data sources had inconsistent quality between markets, with no systematic way to identify and fix failure modes.
Approach
- Designed and operated data pipelines processing millions of competitor data points daily, with deduplication, schema normalisation, and entity resolution across markets.
- Built monitoring and alerting for multiple fraud-related metrics, coordinated with internal stakeholders so the signals stayed actionable rather than noisy.
- Trained and iteratively improved the entity resolution model, with in-depth performance analysis across markets to surface market-specific failure modes and feed findings back into training data, features, and thresholds.
Outcome
- Transformed a system riddled with errors that had little trust from the business into one used daily by executives and local teams.
- Fraud monitoring and competitive intelligence pipelines running in production across multiple markets, feeding both product teams and pricing decisions.
- Double-digit gains in entity resolution quality across iterative training and evaluation cycles.
Have a system that needs to work in production?
Tell us what you are building, or what has stopped working. Thirty minutes, no pitch deck.