Retail Labor Planning Model – Alix Partners Carolyn Taricco Erin Gripp Victoria Cohen
Alix Partners • “AlixPartners is a global firm of senior business and consulting professionals thatspecializes in improving corporate financial and operational performance, executingcorporate turnarounds and providing litigation consulting and forensic accounting services when it really matters – in urgent, high-impact situations.”
Project Description • Develop a model, using real sales and labor data for a U.S. retail chain, that can predict, by store, what the future labor needs are given specific sales targets.
Project Goals • The model should predict future labor needs given specific sales targets • Separate stores into productivity groups given labor, sales, and attributes • Create additional analysis that would be interesting to the end user
Project Objective • Expected approach: run regression analyses to come up with a formula for calculating labor amount given sales amount. The model will need to take in sales, labor, and store attribute data to run an analysis and come up with: 1. Best fit equation or each store allowing someone to input a sales target and get an approximate number of labor needed to meet target 2. For each store, a list of similar performing stores 3. For each store, denote whether any of the attributes are contributors to their labor productivity
Success & Completion Criteria • Develop an applicable model using a different approach then previously used. If time allows, compare to original approach to determine benefit provided.
Project Assumptions • Data provided (attributes, sales data, etc.) • Model needs to be applicable to other chains
Our Solution • Developed a systematic approach that calculates a suggested amount of labor hours for a given store to be considered efficient, relative to the most efficient stores based on the historical labor data we received.
Analysis of Situation: Approaches • Data Envelopment Analysis (DEA) • Approach for evaluating efficiency of individual units (DMUs, in our case Stores) • Efficiency is estimated relative to other units in the sample • Benefits over Regression • Uses a series of optimizations (1 for each DMU) rather than a single optimization for all observations • For each inefficient store, it indicates the efficient reference set • those on the efficient frontier against which the DMU is directly compared
Analysis of Situation: Considerations • Produces a measure of how efficiently inputs are utilized to obtain outputs • The input being labor hours and output being sales dollars • We also took into account 2 other inputs that influence sales: • General Manager Tenure • Turnover
Technical Description • Sorted given data on 519 stores using AWK • Labor hours, sales dollars, attributes • Efficiency Frontiers Generated using DEA • By quarters • 3 Tiers • Constraints • Limit the weight on the 2 attributes to no more than 15% of the inputs
Conclusion & Critique • Identified highly efficient individual stores • Labor Hours and Sales Dollars of these stores • Found that the stores with tier 1 efficiency levels in quarter 1 2007: *$174.10 sales dollars generated per labor hour