OUT OF SHELF DETECTION UPLIFTS
SALES FOR MIDSIZE RETAILER
We have been reached by a supermarket chain that operates over 200 stores and sells more than 5.000 SKUs, with over 20% of Private Labels share. We have analysed the business, and together we have decided to develop the Shelf Video Recognition solution for automated video analytics of products in every store, which helped to:

  • increase by 10% on-shelf availability within the core assortment;
  • detect hourly out of shelf incidents (aka missed sales);
  • enrich the traditional selling history with the shelf pulse for advanced analytics;
Amazon Platform with GPU / Tensorflow / Node.js for operationalization / SAP for data integration
DESCRIPTION
THE CHALLENGE
The retailer had a discount-driven model with the turnover of the highly promoted products. The assortment matrix was kept identical and small across all stores in the chain, keeping the supply chain costs low. This retailer lacked visibility into product availability for some of its key product categories, which could lead up to 2% of yearly losses. While working with Neurons Lab to understand current levels of out of stocks for core products, they found that the average out-of-stock persists for 24 hours before corrective action is taken.
THE SOLUTION
THE RESULT
Our team already had experience with the Image Recognition technologies, which allowed us to launch a pilot project to detect a limited range of products very quickly - within couple of months. Image recognition solutions captured images of retail shelves, which are then processed and analysed to offer insights around Out of Shelf (OOS). This method provided access to real-time data about every SKU in the pilot stores, alerting store staff about the OOS event occurred.
Within two months, client has integrated a Proof of Value solution with the following outcomes:

  • First of all, the retailer managed to digitise losses from "Out of Shelf" incidents with high accuracy, which could not be accurately counted earlier.
  • Secondly, with the help of our solution, it was possible to reduce the length of the process of reacting to the Out of Shelf events from 10-12 hours to 15-minutes.
  • The OOS information made it possible to feed additional features to the existing demand forecasting model, gaining additional accuracy.
REQUEST
AI AT WORK
MEASURED
Ukraien, Kiev
Cupertino str. 132
info@neurons-lab.com
+44 330 027 2146
International House,
64 Nile Str, London, N1 7SR,
United Kingdom
© 2020 Neurons Lab LTD. Registered in England and Wales
Company Number 12265479