Pythian assists global retailer with advanced analytics to increase revenue

“By modernizing our data warehouse with Google Big Query we were able to put a massive volume of in-store behavioral data to work to market to our clients in much more targeted and effective ways.“—Chief Data Architect


Business Needs

The client wanted to be able to maximize floor space revenue by showcasing merchandise in the best possible locations within each store. They also wanted to gain new insights into shopper behavior to promote cross-brand shopping as well as optimize staffing levels to deal with the changes.


Pythian modernized the company’s data warehouse on Google Cloud Platform,  and introduced a cost-effective, real-time data pipeline that ingested and processed a stream of live in-store shopper location data. Data was drawn from two sources - the network of over 50,000 access routers equipped with WiFi that captured beacons from shoppers’ smartphones, and from sales data. This data was analyzed in near real time to identify which retail displays were generating the most traffic. Pythian integrated all the pieces of a modern data warehouse to enable real-time availability and integrity of shopping data. The data warehouse was made up of these components:
  • Databus: Pythian recommended Google Cloud Pub/Sub because of its usage-based billing,  fully-managed, its ability to scale globally, its extensive out-of-the-box monitoring, and because it eliminates the need to manage HW/SW.
  • Data processing: Pythian used Google Cloud DataFlow for its portability, native streaming of data and support for batch processing.
  • Data storage: BigQuery delivered powerful real-time data access and analytics capabilities, support for streaming data and eliminated unnecessary infrastructure and maintenance costs. 
  • Cloud storage: Google Cloud Storage was used for archival of raw data. 
  • Data analysis and visualization - Visualization was built on a Google Maps base.


With the new data platform, the company had access to advanced analytics capabilities and insights they had never before seen. Using the weekly store reports, each store could customize their merchandising based on the people and traffic patterns. Staffing levels were optimized, and the size of purchase gave new information for merchandising. Repeat shopping cycles were leveraged with promotions in-store and online purchases.