The Marble Group: Natural Stone Inventory Intelligence Platform for UK Stone Buyers

Inventory management software development was the missing link for the UK natural stone trade, a market fragmented across dozens of independent wholesalers, each running a completely custom inventory system. Some suppliers run Algolia-powered catalogue search. Others require full browser automation through login walls with infinite scroll. Others expose internal REST or JSON APIs with no public documentation. And others serve classic paginated HTML. Buyers and trade professionals had no way to compare inventory or pricing across suppliers without visiting each site individually and manually reconciling data, a process measured in hours, not minutes. Price lists arrived as Excel files that had to be manually cross-referenced against product pages. There was no single source of truth.

The Marble Group's platform is built around a Django application with a dedicated scraper module for each of the 10 suppliers, all feeding into a single normalised PostgreSQL schema. Each scraper uses the right tool for the job: Playwright with authenticated sessions and infinite scroll detection for Slabware-platform suppliers, async aiohttp with BeautifulSoup for REST/HTML endpoints, and Algolia API integration for catalogue-search-backed suppliers. A shared normalisation layer standardises every incoming record before it touches the database, mapping raw stone type strings and finish names to canonical values via case-normalised lookup tables with 50+ and 60+ entries respectively. A web dashboard exposes the aggregated inventory with filtering. A price reconciliation module accepts uploaded supplier Excel price lists, matches rows by name and thickness, and updates price fields in bulk. PDF and colour-highlighted Excel exports are generated. Discord webhooks deliver scraper run summaries.

Django 5 + Django REST Framework API. PostgreSQL database with Supplier, StoneItemFormatted, StoneItemImage, and SupplierPDF models. Celery for scheduled scraper runs. Playwright (headless Chromium) for authenticated browser automation. async aiohttp with 10 concurrent connections for high-throughput detail fetching. BeautifulSoup HTML parsing. pandas + openpyxl for Excel price list reconciliation. ReportLab for PDF generation. Discord webhook notifications per scraper run.