Transcript Purposes & Risks of Forecasting

Replacing the MPS with a Supply Forecast
to support a Demand Forecast
Jess Marino
December 5, 2012
Contents
 Why forecast?
 Risks and Rewards of forecasting
 Responding to Uncertain Demand
 Supply Forecast
Page 2
2
Purposes & Risks of Forecasting
Forecasts serve many Purposes across the Enterprise
WHAT is done and WHY?
Planning Emphasis and Direction
Region
Product
Line
Revenue Planning
Revenue Scenarios
Hiring
Allocation Criteria
Commissions &
Quotas
Channel
Family
Estimating TAM and Share
Pricing Targets
Programs & Promotions
Margins @ Mixes
Message to Analysts
Product
Customer
Scheduling Factory Volumes
Materials Planning
Balancing Factory Capacity
Capital
Assessing Direct Cost @ Mixes
Analyzing Absorption implications
Business Need / Benefit
Page 4
4
Forecasts are also constructed along Time & Unit dimensions
Time Horizon






Daily
Weekly
Monthly
Annual
Multi-year
Hourly
Unit Of Measure
 Dollars
 Units
Page 5
5
“The” Forecast can mean many things
SEG
SKU
Wk 01
Wk 02
Wk 03
Wk 04
Mo. 1
Wk 05
Wk 06
Wk 07
Wk 08
Mo. 2
Wk 09
Wk 10
Wk 11
Wk 12
Wk 13
Mo. 3
Qtr
A
xxxxxx-xxA
583
522
487
314
1,906
269
257
256
247
1,029
239
249
248
248
145
1,129
4,064
A
xxxxxx-xxB
3,328
4,010
3,992
3,732
15,062
3,372
3,071
2,688
2,638
11,769
2,588
2,746
3,059
2,981
409
11,783
38,614
A
xxxxxx-xxC
16,063 17,474 17,234 18,589
69,360
18,260 18,100 17,710 16,608
70,678
15,507 16,982 16,040 14,428
6,620
69,577
209,615
19,974 22,006 21,713 22,635
86,328
21,901 21,428 20,654 19,493
83,476
18,334 19,977 19,347 17,657
7,174
82,489
252,293
21,457 21,779 21,712 21,581
86,529
21,571 20,188 19,242 19,556
80,557
19,871 19,345 20,111 20,253 20,460 100,040
267,126
Segment A Total
B
xxxxxx-xxD
B
xxxxxx-xxE
434
413
440
450
1,737
B
xxxxxx-xxF
4,604
5,060
5,818
9,764
25,246
B
xxxxxx-xxG
Segment B Total
C
xxxxxx-xxH
C
xxxxxx-xxJ
Segment B Total
2,333
5,919
83,664
162,680
29,261 34,765 40,212 45,896 150,134 46,915 43,759 47,638 49,254 187,566 50,870 50,532 48,440 47,077 37,819 234,738
572,438
55,756 62,017 68,182 77,691 263,646 81,295 76,928 81,211 84,308 323,742 87,405 86,618 86,602 84,761 75,389 420,775
1,008,163
22,938 23,556 21,017 21,225
424
470
480
475
12,385 12,511 13,851 15,023
1,849
53,770
470
469
466
463
465
16,194 16,272 17,585 16,968 16,645
88,736
20,288 19,647 16,701 14,502
71,138
12,304 13,562 13,147 13,183
57,317
217,191
47,266
19,879 20,523 22,077 22,033
84,512
21,988 22,163 22,550 22,728 22,996 112,425
244,203
31,467 33,142 33,549 37,844 136,002 40,167 40,170 38,778 36,535 155,650 34,292 35,725 35,697 35,911 28,117 169,742
461,394
8,529
9,586
12,532 16,619
Grand Total
5,121
1,721,850
Average Selling Price $55.00
Quarterly $ Forecast
$94,701,750
Do we support the weekly SKU forecast?
Do we support the monthly SKU forecast?
Do we support the quarterly SKU forecast?
Do we support the quarterly Segment forecast?
Do we support the quarterly total Unit forecast?
Do we support the quarterly total Dollar forecast?
6
Page 6
6
SKU Forecast accuracy degrades over time
Expected SKU Errors
Over
0
+1
+2
+3
+4 …………………………………………
Time in Future (Weeks)
+n
Under
The further into the future, the harder
to predict details with accuracy
Page 7
7
Unknown Demand creates significant Business Risk
Over-Forecasting
 Excess Finished Goods Inventory
 Excess component and subassembly material
 Distorted priorities
 Over-committing capital
 Reduced prices to unload excess FGI, reducing profits
Potential Under-Forecasting Cost
 Dissatisfied customers
 Reduced revenue
 Premiums paid for expedited materials
 Premiums paid for expedited delivery
The longer the replenishment & fulfillment time,
the greater the risk that is introduced
Page 8
8
Responding to
Uncertain Demand
Operations Leverage on Break Even and ROIC
Work the Numerator and Denominator of
Cash Generated
Cash Invested
 Attack Truly Variable Costs
 Reduce Variability
 Align Production with Revenue
 Focus on Speed and Responsiveness instead of Inventory
 Influence Fixed Costs
 Hiring
 Plant & Property
 Equipment
Page 10
10
Methods to rely less on an accurate forecast
 Plan for Ranges, using statistics and probability – not precise matching
of SKU’s and dates
 Have a stratified plan for different types of Demand
̶ Strategic inventory buffers for specific designs
̶ Planning BOM’s for Assemble to Order
̶ Design for postponement to enable “one to many” configurations
 Establish production processes featuring speed, flexibility, and minimal
variability
̶ TQM, JIT, and Lean Manufacturing
̶ General purpose equipment and common components
̶ Cross training and setup reductions
̶ Leverage Contract Manufacturers for spikes in Demand
Page 11
11
Key Statistic: Theoretical Service Levels
Std Dev.
1
2
3
Theoretical Service Level
84.1%
97.7%
99.9%
Page 12
From technicalchange.com
12
Key Statistic: Process Capability
A process may be “in control” but not
capable. Attempting to support a
service level with an incapable process
will explode inventory without
benefiting customers or profits
From sixsigma.knowledgehills.com
Page 13
13
Key Statistic: Coefficient Of Variation
The coefficient of variation(CV) is a measure of the relative variability of a
process – it is the standard deviation divided by the average.
The coefficient of variation of an arrival process:
Standard deviation of interarriv al time
CVa 
Average interarriv al time
Same Standard Deviation but Different Average
Average = 100 Stdev = 10
CVa = 10 /10 = 1
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
CVa = 10 /100
= 0.1
Inter-arrival time /Average
Inter-arrival time /Average
Average = 10 Stdev = 10
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Observation
* From Cachon, Matching Supply with Demand, Third Edition, McGraw Hill Irwin
Observation
Page 14
14
Extrapolating COV to project Forecast Accuracy
The greater the COV the greater the business Risk
Coefficient of Variation
Probability Actual Demand is
Less than 75% of Fcst Demand
Probability Actual Demand is
Within 25% of Fcst Demand
0.1
0.6%
98.8%
0.3
15.9%
68.3%
0.5
30.9%
38.3%
0.8
36.9%
26.1%
1.0
40.1%
19.7%
1.5
43.4%
13.2%
2.0
45.0%
9.9%
3.0
46.7%
6.6%
* From Cachon, Matching Supply with Demand, Third Edition, McGraw Hill Irwin
Page 15
15
Supply Forecast
Plan to be READY – do not expect to execute the precise plan
Reduce the urge to focus on matching numbers to plans that will never be executed
Product WW35 WW36 WW37 WW38 WW39 WW40 WW41 WW42 WW43 WW44 WW45 WW46 WW47 WW48 WW49 WW50 WW51 WW52 WW01 WW02 WW03 WW04 WW05 WW06 WW07 WW08 WW09 WW10 WW11 WW12 WW13
A
1,448 1,048 1,038 1,243 1,348 1,918 1,999 2,032 1,933 1,751 1,604 1,649 1,977 1,721 1,747 1,691 1,757 1,494 1,429 1,195 1,195 1,195 1,334 1,242 1,273 1,277 1,358 1,358 1,285 1,271 1,251
B
581
800
665
566
270
193
173
209
142
159
186
187
191
213
153
184
179
131
278
268
264
264
264
264
264
252
259
259
259
259
176
C
188
264
308
434
460
509
529
430
677
356
440
517
488
481
482
482
445
405
395
394
394
393
393
393
389
385
384
383
383
322
257
D
758
585
791
620
591
564
246
705
731
694
631
694
473
593
593
593
593
366
526
490
490
490
490
490
490
469
480
480
480
480
80
E
1,787 1,578 1,036 1,545 1,443 1,629 1,469 1,223 2,016 1,702 1,903 2,072 1,932 1,043 1,919 1,673 1,268
236
954
926
926
926
941
926
926
908
915
908
908
839
558
Establish FGI buffers based on Variability
 Lowest COV requires smallest buffer
 Highest COV should not be buffered (discrete orders with Lead Time)
 Moderate COV tied to margins and customer strategic importance
Establish non-FGI buffers based on commonality and annual usage
 Components for Make To Order
 Key subassemblies for Assemble To Order
 Standard semi-finished units for Postponement
Prioritize daily build based on Business Rules
 Orders
 Buffer replenishment
 Other
Page 17
17
Plan to be READY – Plan overall Output & Capacity; not Daily Mix & Qty
• Orders
• Buffer below target
FGI
Avail.
No
Coverage?
in WIP?
Con-
No
No
strained?
Start per
Authorized
• Revised Targets
Pool
Yes
Yes
Yes
Fill
Fill
Prioritize /
from Stock
From WIP
Reconfigure
Major Assumptions
 Quarterly targets established for output, revenue, cost, inventory (S&OP)
 Minimum capacity is established each week (can vary within a range)
Authorized Starts
• Fill Past Due + Current
Week Backlog
 Buffer quantities based on rate and COV
• Replenish Buffer Levels
that are below target
 Component buffer inventory established using Forecast as an input
• Adjust to revised Targets
 Business Rules for executing daily starts (S&OP)
• Linked to component
availability
 Daily Dispatch List tied to business rules and key constraints
• Linked to capacity
utilization target
Plan to Forecast; Build to Demand
Page 18
18
Summary
Maximize TODAY’s Starts and TODAY’s Moves
Supply
Total
Demand
Postponement & Build To Order
Discrete Orders, Buffer replenishment,
Early Build of low COV SKU’s
Standard Configurations for Postponement
Time
19
Page 19
19
Selected Postponement Case Study Slides:
Benetton
Provided by Sara Angela from Monza, Italy
Benetton Strategy & Approach
Knit
Dye
Dye
Knit
POSTPONEMENT
21
Page 21
21
Benetton results from Postponement
Benefit
Profit
a - Lead time reduction
=
Customer satisfaction
b - Smaller unsold stock
=
Less operating costs
c - Wide number of colors
=
d - Less expensive inventories
=
Greater
competitiveness
Less operating costs
e - Platforms design standardization
=
More efficiency
f - Constant feedback on the demand evolution =
Greater competitiveness
a + b + c + d + e + f = more sales, less costs, more competitiveness, more efficiency
MORE REVENUE
22
Page 22
22