Coles Supermarkets Cuts Out-of-Stock Events by 38% with a Real-Time Supply Chain Platform

Challenge

, one of Australia's largest supermarket chains, was grappling with persistent out-of-stock events across its 800+ store network. Inventory data lived in siloed systems across suppliers, distribution centres, and stores — meaning by the time a stock gap was identified, it had already impacted customers for hours or days.

Demand forecasting relied on historical averages that couldn't account for local events, weather patterns, or promotional spikes. The supply chain team spent significant time each week manually chasing suppliers for delivery updates and reconciling discrepancies between expected and actual stock levels.

"Every empty shelf is a lost sale and a disappointed customer," said Coles' Head of Supply Chain Innovation. "We needed to see around corners — to know a gap was coming before it arrived."

Solution

Winklix built a real-time supply chain visibility and demand forecasting platform for Coles, integrating data from 200+ suppliers, 12 distribution centres, and 800+ stores into a unified Snowflake data warehouse, with a predictive ML layer on top. Delivery ran over 13 months.

Data Integration & Warehouse

Winklix built Kafka-based data pipelines to ingest real-time feeds from supplier EDI systems, distribution centre WMS platforms, and in-store POS data. All streams were normalised and loaded into a Snowflake cloud data warehouse, creating a single source of truth for inventory across the entire network.

Predictive Demand Forecasting

A Python ML model incorporated historical sales, local event data, weather forecasts, and promotional calendars to generate SKU-level demand predictions at the store level, 14 days in advance. Automated replenishment recommendations were surfaced directly to category managers.

Operations Dashboard

A React-based dashboard gave supply chain managers a live view of stock levels, inbound deliveries, and predicted stock-out risk across every store and category. Automated alerts notified the relevant buyer or supplier contact the moment a risk threshold was crossed.

Outcome

Within six months of full deployment across the Coles network:

• Out-of-stock events reduced by 38% across all store categories.
• Supplier on-time delivery rate improved by 24% through proactive exception management.
• Food waste reduced by 15% through more accurate demand-driven ordering.
• Manual supply chain reporting effort reduced by over 30 hours per week.
• Forecast accuracy improved from 71% to 89% at the SKU-store level.