![Page 1: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/1.jpg)
S SN
S SN
Switching Rules for Optimal Ordering in a CPG Company
By Shilpa Shenoy and Ai ZhaoAdvisor: Dr. Marina G. MattosMIT SCM
1
![Page 2: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/2.jpg)
S SN
ContentsProduct Overview
Research Problem
Data Analysis
Model Development
Results & Conclusion
2
![Page 3: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/3.jpg)
S SN
Consumer Product Sourced and Manufactured in one
country
Distributed to retail channels across the
world
Product OverviewSENSE is in the FMCG industry with thousands of products
High demand Medium demand Low demandProject sample size:3 SKUs with differing demand patterns
3
![Page 4: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/4.jpg)
S SN
ContentsProduct Overview
Research Problem
Data Analysis
Model Development
Results & Conclusion
4
![Page 5: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/5.jpg)
S SN
High volatility and unexpected spikes in demand
Discounts offered for greater quantities
Varying MOQ~ 1 month of demand
Cannot leverage the discount
High cost and stock-outs1 2 3 4 5 6 7 8 9 10 11 12
MonthsYear 1 Year 2
$0.27
$0.20 $0.17 $0.16
1,000 5,000 10,000 50,000Ordering quantity
Project objectiveTo optimize the raw material ordering policy for SENSE by determining the best minimum order quantity (MOQ) to use.
Demand pattern YoY(monthly final demand, in units)
Quantity discounts(unit price, in $)
5
![Page 6: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/6.jpg)
S SN
How much raw material to order?Balance ordering and holding costs
6
![Page 7: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/7.jpg)
S SNCo
st
Order Size
Holding Cost Ordering Cost Total Cost
Total Cost = Holding Cost + Ordering Cost
Target Service level = 99.3%In a 1-year period, no stock-out event
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43
Units
Weeks
Order Placed Available Inventory
Solution ApproachMaintain a balance between under-stocking and over-stocking.Goal: Choose an ordering policy which fully avoids stock-outs and is the lowest cost
Available Inventory (weekly final inventory, in units)
Cost vs. ordering quantity(cost with changing order size, in $)
7
![Page 8: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/8.jpg)
S SN
ContentsProduct Overview
Research Problem
Data Analysis
Model Development
Results & Conclusion
8
![Page 9: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/9.jpg)
S SN
Input datasetsTwo sets covering demand and inventory
W1 W6 W11 W16 W21 W26 W31 W36 W41 W46 W51
1st forecast 2nd forecast 3rd forecast
Rolling Forecast Evolution(weekly final forecast position, in units)
Long term
020000400006000080000
100000120000140000160000180000200000
Week 1
Week 4
Week 7
Week 10
Week 13
Week 16
Week 19
Week 22
Week 25
Week 28
Week 31
Week 34
Week 37
Week 40
Week 43
Week 46
Week 49
Week 52
Uni
ts
Usage Safety Stock
Historical Inventory Trend(weekly final inventory position, in units)
Short term
9
![Page 10: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/10.jpg)
S SN
Data analysisForecast accuracy, seasonality, and demand distribution
Rolling Forecast Evolution
1 Forecast Accuracy
+2
Rolling Forecast Evolution
Historical Inventory Trend
Seasonality Analysis
3 Rolling Forecast Evolution
Demand Distribution
10
![Page 11: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/11.jpg)
S SN
Data analysisForecast accuracy, seasonality, and demand distribution
• Forecast vs. actual usage to measure forecast quality
• Forecast error =actual usage - forecast
• Mean Absolute Percent Error (MAPE) to measure accuracy
• Compared demand patterns year-over-year
• Calculated seasonality factors for each year
• De-seasonalized data so identify similarities
• Spread of observations around the mean
• Descriptive statistical analysis
• Observations close to mean à Normal distribution
• Variation equal to mean à Poisson distribution
Forecast Accuracy Identify Seasonality Demand Distribution
11
![Page 12: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/12.jpg)
S SN
Forecast accuracyForecast vs. actual usage for the 3 SKUs under consideration
W1 W6 W11 W16 W21 W26 W31 W36 W41
Forecast Actual Usage
W1 W6 W11 W16 W21 W26 W31 W36 W41
Forecast Actual Usage
W1 W6 W11 W16 W21 W26 W31 W36 W41
Forecast Actual Usage
High demand SKUMAPE = 44%(weekly final demand, in units)
Medium demand SKUMAPE = 26%(weekly final demand, in units)
Low demand SKUMAPE = 40%(weekly final demand, in units)
12
![Page 13: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/13.jpg)
S SN
Seasonality analysisNo seasonality identified. Unexpected demand spikes occur, subject to promotions by retailers
0
100000
200000
300000
400000
500000
600000
1 2 3 4 5 6 7 8 9 10 11 12
Uni
ts
Months
Year 1 Year 2
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
1 2 3 4 5 6 7 8 9 10 11 12Months
Factor 1 Factor 2
Demand pattern YoY(monthly final demand, in units)
Seasonality factors for 2 years(monthly seasonality factor)
13
![Page 14: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/14.jpg)
S SN
Demand distributionfor high demand SKU
Demand distribution(15 bins of demand forecast, in units)
Demand Forecast (Units)
[0, 17,512]
(17,512, 35,024]
(35,024, 52,536]
(52,536, 70,048]
(70,048, 87,560]
(87,560, 105,072]
(105,072, 122,584]
(122,584, 140,097]
(140,097, 157,609]
(157,609, 175,121]
(175,121, 192,633]
(192,633, 210,145]
(210,145, 227,657]
(227,657, 245,169]
(245,169, 262,681]
MeanMean = 80,840Standard Deviation = 53,454
14
![Page 15: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/15.jpg)
S SN
Demand distributionfor medium demand SKU
Demand distribution(15 bins of demand forecast, in units)
Mean = 15,277Standard Deviation = 13,158
Demand Forecast (units)
[0, 3,766]
(3,766, 7,532]
(7,532, 11,298]
(11,298, 15,065]
(15,065, 18,831]
(18,831, 22,597]
(22,597, 26,363]
(26,363, 30,129]
(30,129, 33,895]
(33,895, 37,661]
(37,661, 41,427]
(41,427, 45,194]
(45,194, 48,960]
(48,960, 52,726]
(52,726, 56,492]
Mean
15
![Page 16: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/16.jpg)
S SNDemand Forecast (units)
[0, 332](332, 663]
(663, 995](995, 1,327]
(1,327, 1,659](1,659, 1,990]
(1,990, 2,322](2,322, 2,654]
(2,654, 2,986](2,986, 3,317]
(3,317, 3,649](3,649, 3,981]
(3,981, 4,313](4,313, 4,644]
(4,644, 4,976]
Demand distributionfor low demand SKU
Demand distribution(15 bins of demand forecast, in units)
Mean = 816Standard Deviation = 1247 Mean
16
![Page 17: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/17.jpg)
S SN
ContentsProduct Overview
Research Problem
Data Analysis
Model Development
Results & Conclusion
17
![Page 18: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/18.jpg)
S SN
Model SimulationIterations with different values of input parameters to reach the solution
Input Parameters
Ordering Policy 1
Ordering Policy 2
Ordering Policy 3
Ordering Policy N
Stock-out event
Lowest cost
DRP Model Run
Stock-out units
Ordering cost
Holding cost
Solution: No stock-out and low cost
Safety Stock
MOQ
Using a Distribution Requirement Planning (DRP) system: Covering 4 Weeks of demand
18
![Page 19: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/19.jpg)
S SN
Base Model Ordering Policy: How much and when to order
Safety Stock MOQ
Stock-Out
Events
Ordering Cost
Holding Cost Total Cost
262,702 10,000 0 $ 479,111 $ 5,916 $ 485,028262,702 50,000 0 $ 476,942 $ 6,110 $ 483,051
131,351 50,000 0 $ 476,942 $ 4,691 $ 481,63252,540 50,000 1 $ 476,942 $ 3,757 $ 480,699
52,540 100,000 0 $ 476,942 $ 4,051 $ 480,99326,270 50,000 2 $ 476,942 $ 3,578 $ 480,520
26,270 100,000 0 $ 476,942 $ 3,821 $ 480,763
26,270 150,000 2 $ 476,942 $ 4,192 $ 481,13426,270 200,000 0 $ 476,942 $ 4,256 $ 481,198
0
50,000
100,000
150,000
200,000
250,000
300,000
350,000
400,000
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43
Inve
ntor
y -U
nits
Weeks
Order Placed Available Inventory
Simulation Results(by changing Safety Stock and MOQ)
Inventory Position for best scenario(weekly final inventory position, in units)
19
![Page 20: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/20.jpg)
S SN
Switching RuleSwitch MOQ to higher or lower values depending on the demand forecast
Since demand is highly volatile, there are periods of very high demand and very low demand
Having one MOQ throughout the year can lead to over-stocking during the low demand periods
Holding cost can be further reduced if we switch to lower MOQs for the low demand season
Smaller MOQs also mean higher ordering cost. Need to balance the ordering cost and holding cost
Low Demand
MOQ 1
High Demand
MOQ 2
20
![Page 21: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/21.jpg)
S SN
Simulation with switching rulesExperimented with 5 switching rules
Baseline Compare to MOQ
OR
Lower MOQ
Higher MOQ
Switching rule 1 Switching rule 2 Switching rule 3 Switching rule 4 Switching rule 5 21
Current week’s forecast
Upcoming 4th week’s forecast
Average forecast of next 4 weeks
Average forecast for 1 year
Average actual usage for 1 year
![Page 22: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/22.jpg)
S SN
ContentsProduct Overview
Research Problem
Data Analysis
Model Development
Results & Conclusion
22
![Page 23: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/23.jpg)
S SN
Switching rule resultThe switching rule determines the MOQ values, and when to switch to a lower or higher MOQ.
0
50,000
100,000
150,000
200,000
250,000
300,000
350,000
400,000
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43
Units
Weeks
Order Placed with MOQ = 50,000 Order Placed with MOQ = 90,000
Available Inventory
Forecast Average MOQ
59,669 66,038 50,000
57,205 66,038 50,000
101,744 66,038 90,000
112,617 66,038 90,000
67,813 66,038 90,000
Current demand forecast vs. actual usageMOQ1 = 50,000MOQ2 = 90,000
Ordering policy with switching(weekly final inventory, in units)
23
![Page 24: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/24.jpg)
S SN
Model resultsOrdering policy with 5 switching rule choices
Switching Rule Safety Stock MOQ 1 MOQ 2
Stock-Out
Events
Ordering Cost
Holding Cost Total Cost
No Switching 52,540 100,000 100,000 0 $476,942 $4,051 $480,993
Switching Rule 1 52,540 80,000 155,000 0 $476,942 $4,223 $481,165
Switching Rule 2 52,540 100,000 190,000 0 $476,942 $4,437 $481,379
Switching Rule 3 52,540 50,000 90,000 0 $476,942 $3,862 $480,804
Switching Rule 4 52,540 90,000 150,000 0 $476,942 $4,322 $481,264
Switching Rule 5 52,540 53,000 85.000 0 $476,942 $3,876 $480,818
Comparison between different policies for the high demand SKUThe company can choose the best policy from the options
With a holding charge of 7%
Switching Rule 1:Current week’s forecast vs. Average 1 year forecastSwitching Rule 2:Current week’s forecast vs. Average 1 year usageSwitching Rule 3:Upcoming 4th week’s forecast vs. Average 1 year forecastSwitching Rule 4:Upcoming 4th week’s forecast vs. Average 1 year usageSwitching Rule 5:Average next 4 week’s forecast vs. Average 1 year usage
24
![Page 25: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/25.jpg)
S SN
Conclusion
The model can be used to determine the best material ordering policy. It suggests the safety stock and MOQ value to use
There is no switching rule that fits all products and demand patterns. By changing the input datasets, the company can use the best solution
The ordering cost holds a lot of weight in determining the total cost. Due to this, the cost of the switching rule is close to the base policy
This model standardizes the MOQ to use while re-ordering, and can be applied across other products in the company
25
![Page 26: Switching Rules for Optimal Ordering in a CPG Company · Ordering policy with switching (weekly final inventory, in units) 23. S NS Model results Ordering policy with 5 switching](https://reader034.vdocument.in/reader034/viewer/2022042809/5f8dab02ad457f7af31565db/html5/thumbnails/26.jpg)
S SN
Thanks!
26