Big Data – Retail

Challenge ­

A large European Retailer had data marts spread across different countries, hence it was difficult to get a single view of business. Retailers wanted to build single scalable platform so as to

  • Improve store performance through better understanding of the drivers of individual store potential.
  • Ingest new multi­structured data sources like transaction logs, clickstream data.
  • Inventory forecasting as part of store ordering solution.
  • Analytics solution to provide ability to generate reports quickly from diverse sources.
  • Build better multi­channel shopping behaviour of customers.


  • Lambda Architecture with Hadoop and HBase
    • A single view of the business by mapping all diverse data to a single structure
    • Data capture using Apache Kafka
    • Added semi­structured source like transaction logs, clickstream data to analyse with existing data
    • Scalable, Fault-­tolerant architecture
    • Reduced ETL time
  • Market Basket Analysis
    • When an item goes on sale, let retailers know about adjacent products that benefit from a sales increase as well
  • Real time store performance
    • Implemented real­time dashboards using HBase to understand performance of region/country/store
  • Pattern/Behavioural Analysis
    • Analysis of clickstream data and uncommitted transaction logs data to improve customer retention programme, marketing/promotion programme
  • Inventory forecasting
    • Forecasting of products in 24 hours cycle instead of weekly cycle.

Value to Customer

  • Lower total cost of ownership because of single platform
  • Increased processing speed
  • Real­time view of business
  • Increased store performance to baseline.
  • Optimized product assortment mix to store demographics.
  • Facilitates effective customer centric marketing.

Tags: Analytics, Big Data, Migration, Retail