Store Level Insights
Background
A global CPG company engaged Fractal to operationalize an analytics engine designed to provide store recommendations for maximizing investments in ice cream cabinets. The algorithm was originally developed using Python by Fractal data scientists, and needed to be operationalized in the client’s Azure environment.
Approach
Fractal’s data engineers and data scientists re-factored the existing code to run on Azure Databricks, with mount points for automated ingestion of client-supplied data from Azure Data Lake Storage. The output is stored in Azure Blob, with integrated Azure Active Directory authentication to enable secure access by downstream applications.
Solution Framework
End-to-end, data-to-insights implementation on the Azure Cloud:
- Azure DevOps for automated deployment: The solution was scripted entirely using Azure PowerShell scripts and Databricks APIs for automated deployment from Dev to Test to Pre-Production to Production, all at the click of a button.
- Solution built using Azure PaaS services to enable rapid scalability.
Outcome
Fractal created a first-of-its-kind solution, scalable on Azure across multiple countries to deliver actionable, analytical insights and increased ROI to over 13,000 stores with 17,000 new ice cream cabinet recommendations.