3. Contents
Forecasting
What is a forecast?
Why forecast? Putting it Together
Forecast Types How to Forecast
- Forecast Definition - Total forecast components
- Forecast uses - Statistical forecasts
- Judgmental forecasts
- Benefits of doing forecasts
OUTPUT:
- Consequences of not
Volume: Evaluate TOTAL forecast
- Statistical Methods Value: link to volume forecasts
- Creating Judgmental forecasts
Evaluating the Forecast
4. What is a Forecast ?
• Definition:
A forecast is an estimate of volume or value
– for a given SKU (or group of SKUs)
– for a given period of time.
e.g.
• How much Ventolin Inhaler 200 dose will we sell in
February?
• How much will a case of Augmentin 100 mg cost in
November?
• What % of Panadol sales will we spend on
Television Advertising in 2002?
5. What is a Forecast used for?
• Budgeting / Business Planning
• Input to the PMI process
– Inventory Planning
– Distribution Requirements Planning
– Inventory Management
A Forecast is an ESTIMATE of future sales
It will always be wrong to some extent,
but will always be BETTER than NO information
6. Why forecast….?
• Develop forward view of business
• Allow more informed decision taking
• Highlight gaps vs. budgets
– volume & financial
• Communicate through the supply chain
• Aim for lower costs
– inventory
– write-offs
• PLAN don’t REACT
7. Consequences of not Forecasting
• Always reacting to surprises
– fire-fighting vs. value adding
– no communication through supply chain
• Stock unlikely to absorb abnormal demand
– high stock (wrong stock)
– poor customer service
– unstable NR’s and poor supply
• Forrester Effect
– demand spikes are amplified
• caused by over-reaction to surprises
8. Developing a forecast
- Roles
• Marketing
– Long Term Forecast
• Sales
– Short Term Forecast
• Finance
– Prices & Cost forecasting
• Senior Management
– Sign Off
• Demand
– Process champions
– customer of the forecast
A forecast should be reached by CONSENSUS
9. Monthly Planning Cycle
Forecast
Review
Meeting
Fr Mo Wk. 2
Wk. 1 Th Tu
Finalise
We
Download and We Forecast
KPI
Review Actuals REPORTING
Sales &
Tu
DEADLINE
Th
Marketing
Meeting
Transmit
Monthly
Mo
Demand Meeting
Agreed
Fr
Supply
Plan
Planning
Financial Transmit
Mo
Month Close Cycle
Fr
Net Req’s
Rough Cut
Capacity
Negotiate Planning
Supply
Tu
Exceptions
Th
Review
Global
Meeting
Demand
We
We
Meeting
Tu Th
Global Mo Fr
Supply
Wk. 4 Meeting
Wk. 3
10. Developing a Forecast
- Forecast types
• Forecasts consist of two types of information
– Statistical forecasts
• based on historical data & patterns
– Judgmental forecasts
• based on judgement, research, consensus,
assumptions
11. Developing a Forecast
- Forecast types
Statistical Judgmental
• Base volume • Adjustments
– seasonality – changing market conditions
– repetitive orders – seasonal pattern changes
• yearly tenders – sales promotions
• samples • (with no historical data)
– promotions – random tenders
• (with historical data)
• New Products
• Any situation where historical
information is available and • Any situation where no historic data
reliable. exists or is NOT VALID
12. Developing a Forecast
Forecast Types - SA Investigator
• In SA various volume facts exist to develop the
TOTAL VOLUME FORECAST
– Base Volume • Statistical
– Adjustment Volume • Judgmental/Statistical
– Samples Volume • Judgmental/Statistical
– Free Goods Volume • Judgmental
– Tenders Volume • Judgmental
– TOTAL VOLUME • CALCULATED
13. Developing a Forecast
- Process
• What are the processes that create the different
types of forecasting?
– Statistical
• Base volume forecast
– Judgmental
• adjustments etc.
14. Statistical Forecasting
- Process
• Capture actuals
– from Sales Order Processing (SOP) system
• Filter history
– to remove abnormal demand
– to remove stock outs
– to adjust for step changes
• one off task - when conditions change
• Run forecasting “algorithm”
• tournament, regression etc.
• Evaluate results against assumptions
15. Statistical Forecasting Process
- Capture Actuals
• Actuals from SOP system
– provides historical data to use for statistical method
• essential to drive forecast in future
– will be invoiced sales - therefore :
• all sales will be included
– including promotions volumes
• stock problems will be reflected in lower figures
– requires maintenance to be effective
• initial one off job when first building forecast
• ongoing task is to maintain last month only
16. Statistical Forecasting Process -
Actuals being used as History
• Projects historical sales patterns into future
forecasts
– shape will be ‘smoothed’ to varying degree
160
140
120
100
80
60
40 History
Forecast
20
0
16
19
25
28
43
1
4
7
10
13
22
31
34
37
40
46
17. Statistical Forecasting
- Uses of History
• Trends
250
200
150
100
50
Trending
0
Jul
Jul
Jul
Jul
Jan
Jan
Jan
Jan
Apr
Apr
Apr
Apr
Oct
Oct
Oct
Oct
History Forecast Linear (History)
18. Statistical Forecasting
- Uses of History
• Seasonality
180
160
140
120
100
80
60
40
20
Seasonality
0
1
4
7
13
16
19
22
25
28
31
34
37
40
43
46
10
History Forecast Linear (History)
19. Statistical Forecasting Process
- Filtering (modify) History
• Why Filter
– large abnormal patterns will wrongly influence future
300 Total Vol FC
Sales Vol FC
Sales Vol
250
200
150
100
50
0
Jan
Jan
Jan
Oct
Oct
Oct
Oct
Jan
Jul
Jul
Jul
Jul
Apr
Apr
Apr
Apr
• Things to look for in History to filter
– stock outs (zeros) & promotions (spikes)
20. Statistical Forecasting Process -
Filtering (modify) History
• Results of filtering
– smoother pattern is projected into future
300
Total Vol FC
Adjstmt FC
Sales Vol Md / Fc
250 Sales Vol FC
Sales Vol
200
150
100
50
0
Apr
Apr
Apr
Apr
Oct
Oct
Oct
Oct
Jan
Jan
Jan
Jan
Jul
Jul
Jul
Jul
21. Business Forecasting Process
Capture
Sales/Marketing Historic Data
Responsibility Modify
History
Generate
Forecasts
Review Commercial
Plans
Review Exceptional
Demand
Demand Review
Meeting Sign off
Feed to
Demand Planning
Process
Consensus Forecast
22. Developing a Forecast
- Judgmental Forecasting
• When Statistics won’t work….
– Where there is no reliable history
• New SKUs
– For future events that have no past information
• range changes
• changing market conditions
• promotions
• Solution…….
– Use Judgmental forecasts
• to create forecasts where no statistical can exist
• to adjust statistical volume (as per last slide)
23. Judgmental Forecasting
• Used to add future events to the forecast
– can be positive or negative
– adjustments made to the statistical base
• if one exists (e.g. New SKUs)
• Require assumptions to base judgement on
– research
– market information
– brand plans
– consensus forecasts (Demand Review
Meeting)
24. Judgmental Forecasting
• Adjustments should NOT overwrite base volume
– should be complementary to the statistical
numbers
• statistical Base added to judgement
adjustments
• get the Correct TOTAL VOLUME
– in SA use different Fact for adjustments
• allows analysis & visibility
• comments database can be used to store
assumptions
25. Judgmental Forecasting
300
250
200
150
100
50
0
Jan Feb Apr M ay Jul Aug Sep Nov Dec Feb M ar M ay Jun Jul Sep Oct Dec
Adjstmt Vol 0 0 0 0 33 104 5 0 0 10 20 80 50 40 40 40 40
Sales Vol FC 100 123 131 144 122 80 110 85 112 123 108 144 122 122 110 130 112
Total Forecast 100 123 131 144 155 184 115 85 112 133 128 224 172 162 150 170 152
26. Putting the Forecast together… -
Effort of Forecasting
• Focus Forecasting Effort
– Statistical forecast alone will often achieve sufficient level of
accuracy (especially Cat B/C)
– not always the best solution alone
• Build judgmental adjustments in where necessary
– Complex Demand or High Value (Cat A) Products
• Together powerful tool to deliver
TOTAL forecast
• Statistical - deliver base, trends & seasonality
• Judgmental - promotions, ranges changes,
abnormal scenarios
27. Putting the Forecast together… -
Effort of Forecasting
Value
A
Judgmental
Forecasts
B
Statistical
C Forecasting
Complexity
28. Putting the Forecast together…
- The bigger picture
300
Sales Vol FC Adjstmt Vol
Sales Vol Total Forecast
Sales Vol MD
250
200
150
100
50
0
Jan
Jan
Jan
Jan
Jul
Jul
Jul
Jul
Apr
Apr
Apr
Apr
Oct
Oct
Oct
Oct
29. Putting the Forecast together…
- The Volume - Value link
• Value Forecasts driven by volume
– Volume x Average Price = Sales Value
• both volume and price require forecasting
• Volume - Value link will deliver the
24 month rolling business forecast
– new products will need to be forecast earlier
to get full business picture
• Financial forecasting was deployed in 2000
– will allow forecasting for profit and contribution
• ONE SET of NUMBERS drives the business
30. Putting the Forecast together…
- Points to remember….
• Always remember….
– the forecast from the system may be correct
– the initial assumptions may be wrong
• changing market conditions
• unexpected seasonal conditions
• Work out if adjustments are
– required, realistic and reasonable to make
– if they are - don’t overtype the base volume
• use judgmental facts (i.e. Adjustment vol.)
Notes de l'éditeur
Use this as an ice breaker session: Get flipchart and ask them for their thoughts on why forecasting should happen. Then show the slide and see what they have added extra.