E-commerce · Data · Automation
Prism Price Intelligence
Automated price monitoring for major Hungarian retailers — twice daily, validated, health-tracked, bot-resilient.
3
Retailers integrated
2×/day
Scheduled price runs
6
Validation steps per page
1
Command Docker deploy
The challenge
Manual competitor price tracking does not scale. Major Hungarian shops use bot protection, JavaScript-rendered prices and frequently change page structure — one wrong price is worse than a missing one.
What we built
A modular scraping platform that discovers products via sitemaps, fetches pages through HTTP or headless browser depending on the shop, runs a mandatory validation chain on every page, and stores price history and run health in MongoDB — scheduled twice daily via BullMQ.
Key capabilities
- Three live retailer adapters (VoltMart, MegaStore, TechZone) — modular architecture scalable to 40+ shops.
- Six-step validation chain per product page: HTTP status, soft-block, captcha, DOM, price extraction, anomaly check.
- BullMQ per-supplier fan-out — one shop failure does not block the others; retry-idempotent run IDs.
- REST Health Board API with per-supplier run history (healthy / degraded / failed).
- Docker Compose stack: MongoDB + Redis + engine + web — one-command deploy.
Tech stack
TypeScript · Playwright · Cheerio · MongoDB · Redis · BullMQ · Next.js 15 · Docker · Zod