Mar 27, 2017 - Where and why do you use forecasting and analytics in your business? Long term / Strategic (+3y). Finance
REAL IMPORVEMENTS TO YOUR SUPPLY CHAIN!
Best practice in Analytics and Forecasting (e.g. for managers supervising the job) Jonas HATEM (+44 7530 12 77 55 or +32 499 539 713)
March 27th 2017
www.mobiusuk.co.uk www.mobius.eu
About me & MÖBIUS Quick facts 176 Consultants
23 M£ Expected Consolidated Turnover in 2017
4 Countries (+ boots on the ground in a couple more) Over 125 projects/year
[email protected] @jrhatem www.linkedin.com/in/jonasranihatem +44 7530 12 77 55 & +32 499 539 713 +44 2033 974 044
p. 2
Where and why do you use forecasting and analytics in your business? Long term / Strategic (+3y) Finance. Equipment/equipment replacement needs, timing and amount of funding/borrowing needs.
Medium term / Tactical (3m to 3y)
Human resources. Organisational development, staff Accounting. New product/process cost estimates, capabilities planning. profit projections, Budgeting. Marketing. Branding Finance. Equipment/equipment replacement needs, MIS. New/revised information systems; Internet timing and amount of funding/borrowing needs. services. Human resources. Hiring activities, including Operations. NPI planning, capacity planning, recruitment, interviewing, training, layoff planning. network design Marketing. Pricing and promotion, e-business Product/service design. Revision of current features, strategies, global competition strategies. design of new products or services. MIS. New/revised information systems; Internet services.
Short term (8-13w, up to 1y)
Everywhere Operations. Workloads, inventory planning, make or-buy decisions, outsourcing. Sales & Operations planning Product design. Revision features, New prod./services
3
Accounting. Cost estimates. Finance. Cash management Human resources. Hiring activities, including recruitment, interviewing, training, layoff planning, including outplacement, counselling. Marketing. Pricing and promotion Operations. Procurement, Production Schedules, work assignments and workloads, inventory planning.
Where and why do you use forecasting and analytics in your business?
to reduce
variability in the organisation (and make more money with less stress)
4
Where and why do you use forecasting and analytics in your business? ▪ But…
Deviation from average
Average demand
Demand
… it can go terribly wrong
5
5 Best Practices in Analytics and Forecasting
Understand the lingo, data & questions Forecast at the appropriate level Measure your performance Know your cost of forecast mistakes Incentivising on forecast accuracy is usually a bad idea
6
Understand the lingo, data & questions ▪ Get the lingo right – Small data vs. Big data (vs. Cloud) – Descriptive vs. Predictive vs. Prescriptive vs. Automated Analytics
▪ Understand your (Historical) Data Management practices
▪ Frame your questions, don’t go digging.
7
5 Best Practices in Analytics and Forecasting
Understand the lingo, data & questions Forecast at the appropriate level Measure your performance Know your cost of forecast mistakes Incentivising on forecast accuracy is usually a bad idea
8
Forecast at the appropriate level ▪ Forecast planning levels & horizons
Forecast Planning Levels and Horizons Aggregation Business Level
St rat eg ic
Vo lum e Level Mix Level
Tact ical Op erat io nal Frozen
Flexib le
Free
Tho ug h g enerat ed , m aint ained , and used at d ifferent levels o f ag g reg at io n and ho rizo ns, p lanning levels sho uld b eco m e co m m unicat ing vessels and p asst hro ug h info rm at io n fro m o ne level t o ano t her Disag g reg at e hig her-level fo recast s int o m o re d et ailed levels fit fo r use in m o re sho rt -t erm ho rizo ns 9
t
Understand in which stability zone you are
Firm/Frozen zone
Trading Zone / Flex
Free Change zone
Flexibility Commitment cost
Customer Commitment (cust. order)
Commitment Commitment Commitment Little or no To To To Commitment Distribution Production Supplier
Order fulfillment
Distribution Production decision decision manage Demand to meet Supply
Purchase Time decision manage Supply to meet Demand
10
5 Best Practices in Analytics and Forecasting
Understand the lingo, data & questions Forecast at the appropriate level Measure your performance Know your cost of forecast mistakes Incentivising on forecast accuracy is usually a bad idea
11
Ever encountered one of these? ▪ How do you measure forecast performance?
12
Measuring Forecast Performance: issues 1 and 2
timing bias
magnitude Demand Forecast
13
Measuring Forecast Performance: issues 3 and 4 ▪ Mix/aggregation hides forecast mistakes A
B
Aggregated (A+B)
Forecast
75
25
100
Actual
25
75
100
Accuracy% = Max (0 ; 1 - Error %)
0% 0% 100% Max (0 ; 1- |25 – 75| / 25) Max (0 ; 1- |25 – 75| / 25) Max (0;1-|100–100|/100)
▪ Timing aggregation mistakes hides forecast mistakes Product A
Jan
Feb
Mar
Apr
Ma
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Yearly
Forecast
100
0
100
0
100
0
100
0
100
0
100
0
600
0
100
0
100
0
100
0
100
0
100
0
100
600
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
100%
Actual
Accuracy%
Measuring Forecast Performance ▪ No one number will tell the whole story.
Metric
Good For
Mean Square Error (MSE)
Gives greater weight to larger deviations (which could result from outliers)
– Choose your metric according your use, e.g. – Newer approaches tend to combine 2 or more indicators (magnitude and bias), e.g.
Mean Absolute Deviation (MAD)
Use to see magnitude of effect on production – volume dependent
Can’t compare across products; doesn’t indicate if fcst too high or too low
Mean Absolute Percent Error (MAPE)
Independent of product volume – good for products sold consistently
Doesn’t indicate if forecast too high or too low; doesn’t indicate magnitude of impact on production
Normalized Forecast Error
Varies between 1 and –1. Shows whether consistently too high or too low
Independent of volume; doesn’t show magnitude of production impact.
• Mean Absolute Scaled Error (magnitude): MASE • Used together with Accumulated Forecast to Actual Ratio (bias): AFAR
▪ Measure accuracy at the level at which you forecast or higher.
Limitations
▪ Aggregation distorts accuracy measures ▪ Beware of forecast accuracy metrics and intermittent demand
15
5 Best Practices in Analytics and Forecasting
Understand the lingo, data & questions Forecast at the appropriate level Measure your performance Know your cost of forecast mistakes Incentivising on forecast accuracy is usually a bad idea
16
Know your cost of forecast mistakes ▪ Cost of over forecast
▪ Value add of forecasts
– Inventory – Service
Can be calculated
– Capacity – etc…
▪ Cost of under forecast – Cost of lost sales? Can be estimated
17
▪ Why/How should you use this?
5 Best Practices in Analytics and Forecasting
Understand the lingo, data & questions Forecast at the appropriate level Measure your performance Know your cost of forecast mistakes Incentivising on forecast accuracy is usually a bad idea
18
Incentivising staff on forecast accuracy is usually a bad idea Improve FCT by 1%
"N OR MA L " A B C PATTER N
A B C
0
1
scenario 2
80% 60% 35%
81% 60% 35%
80 61% 35%
80% 60% 36%
average accuracy
58.3%
58.7%
58.7%
58.7%
margin contrib. weighted accuracy volume weighted accuracy
12.6%
14.0%
13.9%
13.8%
66.3%
70.1%
69.8%
69.6%
15%
3
50% 35%
71%
35% 50% 20% 15% SKU%
9%
VOL UME%
Classification
SKU%
Volume% Agv.Margin
A
15%
50%
25%
B
35%
35%
C
50%
15%
Focus on A!
MARGI N%
Turnover
Margin
Margin%
FCT accuracy
Inventory
New Invent.
Difference
£ 25,000 £ 6,250
71%
80%
£ 1,500
£ 1,425
£ 75
10%
£ 17,500 £ 1,750
20%
60%
£ 3,500
£ 3,413
£ 88
10%
£
9%
35%
£ 5,000
£ 4,923
£ 77
7,500 £ 50K
750 17.5%
10K = 5 turns
SS = z * σ * D
Focus on B!
終 [end]
20
ありがとうございます [Many thanks]