Retail Inventory Allocation Optimizer

A tool that determines the profit-optimal inventory level for a seasonal item

David Leonard

Allocating seasonal retail goods to stores is a balancing act

  • Seasonal retail products have a short life and generally a limited supply (think fashion apparel, for example)
  • Goods are acquired prior to the start of the season, and held at a Distribution Center (DC)
  • In-season, units are allocated out of the DC to stores on an as-needed basis
  • How do we decide how many units to send to each store in each allocation cycle?
    • If we stock too many, we could end up with "stranded" inventory at the end of the season; if we can't return those to the vendor, then we take a loss on the cost of each stranded unit.
    • If we stock too few, we will lose out on sales during the next "coverage period"
    • The challenge: find the level of inventory that balances those two risks

We can use probability theory to estimate the total costs at different stocking levels

  • A gamma distribution is typically used to describe retail consumer demand
  • Given the demand rate and variability of demand, we can determine the probability of selling a given number of units over a specified period of time
  • For allocation, we are concerned with two time periods:
    • The next coverage period (time between arrival of shipments plus the transit time of an order from the DC to the store), and
    • The time remaining in the season
  • The total cost associated with a stocking level s is the sum of:
    • Lost margin = pr(Coverage Period Demand > s) x (Unit Margin)
    • Obsolescence cost = pr(Remaining Season Demand < s) x (Unit Cost) x (Obsolescence %)
  • Compute cost for each stocking level below cumulative probability <99.9%, then find the level with the lowest total cost

The Optimizer Finds the Most Profitable Inventory Level Based On Input Parameters

Example: Daily Demand = 1 unit, Coverage Period = 9 days, Remaining Season = 4 wks, Variability = 40%, Margin = 50%

plot of chunk unnamed-chunk-1

The Inventory Optimizer Is Available For Evaluation

  • You can try it here
  • Go to the Help tab for definitions of the inputs and other notes
  • Use the sliders to adjust input parameters and observe the effect on the optimal inventory level
  • Ideas for future enhancements:
    • Support additional inputs, such as
      • coverage period and remainder of season average daily demand
      • Item price or cost
      • Obsolescence cost (as % of item cost) - currently it assumes 100%
    • Integrate with forecasting system to directly access demand predictions and measure of variance