A unified workspace for assortment
planning
Client overview
A leading US book distributor and merchandiser plans assortments across retailers, events, product families, planograms, stores, and selling periods. Its assortment-planning teams rely on assortment analysis to make add, drop, and keep decisions on titles across events and retailers.
Challenges
Each event review began with the same manual groundwork. Merchandising teams moved between Blue Yonder, Power BI, and Circana, and reconciled competitor, forecast, planogram, and Title Pool data through manual, multi-sheet Excel lookups. Existing titles (actual performance) and new-item candidates (forecast and industry measures) were evaluated in separate files.
Each time the event, retailer, or reporting context changed, similar reports were rebuilt, with columns repeatedly selected and reordered; and separate templates were maintained for different event types. Facings across outputs also needed rework before add, drop, and keep analysis could begin; and few users understood the underlying workbook logic. This spreadsheet-heavy process increased the risk of manual errors and left analysts with less time for the merchandising decisions their work was meant to inform.
Solution
Accelerating assortment planning through a unified workspace
Zensar built a Streamlit application on Databricks that consolidates assortment data and calculations into a single workspace for existing titles and new-item candidates. Calculations are repeatable and traceable to Databricks sources, and the solution is estimated to reduce manual report-preparation effort by approximately 70%.
Users can select the event name, week filter, and title pool DB key in a single interface and export the active view to XLSX or CSV for continued Excel review, preserving familiar facings across output.
A single view of assortment data: Sales, inventory, planogram, forecast, title pool, and Circana data is consolidated in Databricks, so users no longer need to look across systems.
Right measures for every title: Existing titles are evaluated on actual POS, facings, inventory, and dollars per facing; new-item candidates use the Blue Yonder forecast, 10-Week SPS, MSRP, and Circana data. Calculations are repeatable and traceable to Databricks sources.
One interface for all use cases: A single Streamlit interface, driven by the event name, week filter, and title pool DB key, replaces event-specific templates. The Manage Columns and Save Views options let users add, hide, reorder, and bookmark columns across 100+ fields.
Familiar Excel output, with less rebuilding: Title pool candidates are appended to the same facings across output, with planograms sorted by linear footage and UPCs ranked on dollars per facing. Active-context XLSX and CSV exports support continued Excel review, while purpose-specific data views, on-demand fields, caching, and row limits help manage wide, high-volume data.
Solution enablers
Tech Stack
Streamlit
Databricks
Python and SQL
XLSX/CSV exports
Zensar Services
Product consulting
Data engineering
UI engineering
Reporting and analytics engineering
AWS Services
Azure Databricks
Business outcomes
Less time rebuilding reports. More time making decisions.
Assortment-planning teams now evaluate existing titles and new-item candidates in a single workspace, with less reliance on manually assembled spreadsheets and the familiar Excel output retained. Analysts have more time for strategic add, drop, and keep decisions; and the platform provides a scalable foundation for additional retailers, KPIs, and data sources.
~70% estimated reduction in manual report-preparation effort*
Reduced reliance on manually assembled spreadsheets
Consistent, traceable calculations linked to Databricks sources
More time available for add, drop, and keep decisions
