Measurement and Evaluation of Retail Promotion Authors: Asen Kalenderski and Satya Sanivarapu Sponsor/Advisor: ProdCo Advisor: Dr. Chris Caplice, Dr. Francisco Jauffred MIT SCM Research Fest May 21, 2015
Agenda MIT SCM ResearchFest May 21, Importance of Promotions Promotion Performance Cuboid Case Study: Classification of Dataset Key Insights Questions
Why are promotions important? MIT SCM ResearchFest May 21, Retail sales in March 2015 were $441.4 billion Large retailers spend 10 to 20 percent of their sales on promotions Spending helps increase sales by 30 percent Only 18 percent of the promoted brands create increased store profits
What makes promotions difficult? MIT SCM ResearchFest May 21, When? What? Where? Inventory? Stock-outs? Forecast? Sales? How? Why? pricing supply chain collaboration retailers advertising strategy planning production execution capacities replenishments
Promotion characteristics Promotion Period Inventory ramp up replenishment
Pillars of evaluation MIT SCM ResearchFest May 21, Days of supplyStock-outsSales accuracy to forecast
Days of Supply MIT SCM ResearchFest May 21, Lower Boundary of Bins for Days of Supply at the End of Promotion
Days of Supply Difference MIT SCM ResearchFest May 21, Lower Boundary of Bins for Days of Supply Difference
Days of Supply Difference Percentage MIT SCM ResearchFest May 21, Lower Boundary of Bins for Days of Supply Difference %
Sales accuracy to forecast MIT SCM ResearchFest May 21, Lower Boundary of Bins Sales accuracy to forecast
Promotion Metrics MIT SCM ResearchFest May 21, Sales accuracy to forecast HighSales = ForecastLow Stock-outs HighLow Days of Supply HighGreen ZoneLow
Agenda MIT SCM ResearchFest May 21, Importance of Promotions Promotion Performance Cuboid Case Study: Classification of Dataset Key Insights Questions
Promotion Performance Cuboid MIT SCM ResearchFest May 21, DoS Difference Low Green Zone High Sales Accuracy Negative Sales=Forecast Positive 0 +20% -20% <-20% >20% 0 > 0 < 0
Agenda MIT SCM ResearchFest May 21, Importance of Promotions Promotion Performance Cuboid Case Study: Classification of Dataset Key Insights Questions
Case Study: Supply Chain MIT SCM ResearchFest May 21,
Case Study: Classification (Low SO) MIT SCM ResearchFest May 21, % 4.44% 13.24% 12.93% 7.88% 10.70% DoS Difference Low Green Zone High Sales Accuracy Negative Sales=Forecast Positive 7.10% 7.86% 8.15% 0 +20% -20% <-20% >20% 0 > 0 < 0
Case Study: Classification (High SO) MIT SCM ResearchFest May 21, % 0.01% 0.07% 0.08% 0.07% 2.58% 0.34% 3.44% 0.02% DoS Difference Low Green Zone High Sales Accuracy Negative Sales=Forecast Positive 0 +20% -20% <-20% >20% 0 > 0 < 0
Case Study: Classification (Major Cubes) MIT SCM ResearchFest May 21, A 21.03% B 13.24% C 12.93% D 10.70% DoS Difference Low Green Zone High Sales Accuracy Negative Sales=Forecast Positive 0 +20% -20% <-20% >20% 0 > 0 < 0
Case Study: Classification (Major Cubes) MIT SCM ResearchFest May 21, A 21.03% SKU Sales – Low Replenishments – On time Inventory at store – Low
Case Study: Classification (Major Cubes) MIT SCM ResearchFest May 21, B 13.24% SKU Sales – High Replenishments – On time Inventory at store – Just right
Case Study: Classification (Major Cubes) MIT SCM ResearchFest May 21, C 12.93% SKU Sales – Low Replenishments – On time Inventory at store – Just right
Case Study: Classification (Major Cubes) MIT SCM ResearchFest May 21, D 10.70% SKU Sales – Low Replenishments – On time Inventory at store – High
Agenda MIT SCM ResearchFest May 21, Importance of Promotions Promotion Performance Cuboid Case Study: Classification of Dataset Key Insights Questions
Key Insights MIT SCM ResearchFest May 21, Days of Supply Difference High Look into Replens frequency and quantities Low High Inventories Green Zone High Look into Replens frequency and quantities Low SKU Sales strategy Low High Inventories Low SKU Sales strategy Focus Areas Stock-Out
Agenda MIT SCM ResearchFest May 21, Importance of Promotions Promotion Performance Cuboid Case Study: Classification of Dataset Key Insights Questions
MIT SCM ResearchFest May 21, DoS Difference Low Green Zone High Sales Accuracy Negative Sales=Forecast Positive 0 +20% -20% <-20% >20% 0 > 0 < 0