2. <Insert Picture Here>
Come non impazzire nel gestire la pianificazione della Domanda
Milano, Caffe Panzera, 23 Marzo 2010
Paolo Prandini
Master Principal Sales Consultant, Supply Chain
3. Da sempre l’uomo cerca di prevedere il futuro
Strumenti di
previsione
Dadi
Maghi
Tarocchi
Fondi di Caffè
4. Da sempre l’uomo cerca di prevedere il futuro
Strumenti di
previsione
5. Perche’ la previsione della domanda è importante?
Costituisce la base dei piani di approvigionamento
Costituisce la base per i piani di produzione
Costituisce la base per i piani di Budget
Aiuta a ridurre le scorte
Aiuta ad aumentare la soddisfazione del cliente
Aiuta a massimizzare il ROI promozionale
E‟ alla base delle strategie dei Business „Demand
Driven‟
7. il Demand Management è un processo incompreso
Il motore statistico spaventa
Si pensa servano statistici
esperti in camice bianco
Si pensa sia complicato ed
oneroso da gestire
La figura del Demand Planner
puro è rara da trovare in
azienda e fuori
Ci si affida spesso a quanto
presente a livello di ERP ma
poi non è abbastanza...
8. Ha le sue strane parole chiave...
MAPE (Mean Absolute
Percentage Error)
MAD (Mean Absolute
Deviation)
SD (Standard Deviation)
Accuracy
Bias
Absolute Error
Baseline
Uplift
9. <Insert Picture Here>
Domanda Indipendente,
Prof. Jacobas, Univ.Ilinois
“This demand is primarily influenced by factors
outside the company‟s decisions.These external
factors induce random variation in the demand for
such items, thus demand will be projections of
historical patterns. These forecasts estimate the
average usage rate and a pattern of random
variation
10. Tipi di Demand
Pianificazione Prodotti Configurabili
Domanda Dipendente
12. <Insert Picture Here>
Baseline Forecasting,
definizione
“Baseline Forecasting is a methodology
that uses system inputs and the forecast
engine to develop a statistical plan that may
be further adjusted as needed to provide a
common starting point (or „baseline‟ ) for
internal and external collaboration in order
to reduce forecast error”
13. L’errore sul Forecast produce vari effetti
Forecast Error
Over Forecast Under Forecast
Excess Inventory Order Expediting Cost
Inventory Holding Cost Higher Product Cost
Trans-shipment Cost Lost Revenue
Obsolescence Lost Companion Product Sales
Reduced Margin Lower Customer Satisfaction
14. Esempio
Costs and Lost Sales Example from Forecast Error
Forecast too high:
Monthly SKU Volume 1,000,000 units
Percent Forecast Error 10% Yields: 100,000 units more than required
Average SKU Cost $0.75
Excess Inventory $ per Month $75,000
Annual Excess Inventory $ $900,000
Forecast too Low
Monthly SKU Volume 1,000,000 units
Percent Forecast Error 10% Yields: 100,000 units of lost sales
Average Margin per SKU $0.50
Lost Profit per Month $50,000
Annual Profit Loss $600,000
19. Il Tool ideale quindi deve essere...
Facile da usare
Basato su conoscenze di Business
Che non necessita conoscenze statistiche
Di facile Implementazione e Manutenzione
Con una maggior sensibilita‟...
Con miglior accuratezza...
Che posso gestire in azienda come le altre
applicazioni...
21. Demantra and Supply Chain
Suppliers Finished
Growers
Raw
Manufacturers Product Mfgr
Materials
Brokers
Distributors
Trade Promotion
Retailers
S&OP
Demand Mgmt Business
Distribution
Channels Web Direct Consumer Customers
CONFIDENTIAL: All capabilities and dates are for planning purposes only and may not be used in any contract
22. 0
1
2
3
4
5
6
7
8
Gen 09
Feb 09
Marzo 09
Apr 09
Mag 09
Giu 09
Lug 09
Ago 09
Sett 09
Ott 09
Sales Forecast
Nov 09
Dic 09
Gen 10
La storia da sola non basta per fare previsioni
Sales
23. E’ necessario integrare con eventi Business
Evento Promo Evento
Promo
Sales Forecast (Futuro)
8
7
6
5
4
3
Sales
2
1
0
Mag 09
Feb 09
Lug 09
Dic 09
Marzo 09
Sett 09
Gen 09
Apr 09
Giu 09
Ott 09
Nov 09
Gen 10
Ago 09
24. E’ necessario integrare con eventi Business
Evento Promo Evento
Promo
Sales Forecast (Futuro)
8
7
6
5
4
3
Sales
2
1
0
Mag 09
Feb 09
Lug 09
Dic 09
Marzo 09
Sett 09
Gen 09
Apr 09
Giu 09
Ott 09
Nov 09
Gen 10
Ago 09
27. Demantra quindi...
Puo‟ usare tutte le informazioni che avete circa le
vendite:
Ordinato / Spedito
Calendari Marketing
Eventi Promo
Eventi Media
Syndacated Data (Ac Nielsen, Information Resource)
Dati Demografici
Attributi Prodotto / Store
E restituirvi un Forecast sempre aggiornato e
accurato
In alcuni casi addirittura In tempo quasi reale!*
* Dipende dalla disponibilita’ dei dati e dal tempo di elaborazione
32. Case Study - Fairfax
Anatomy of a Win
Anatomy of a Happy Customer
33. Fairfax Overview
Fairfax is Australia‟s and New Zealand‟s largest
publishing group
(Sydney Morning Herald, The Sun-Herald, The Age,…)
Challenge
Improve demand forecasting (at the kiosk-level)
Improve supply allocation
Reduce lost revenue from sell-outs
Reduce returned copies
34. Oracle Solution
Demantra Value
Scalable forecasting at the most granular level
Effective management of a perishable product (short shelf
life)
More frequent calculation of outlet supply quantities
Solution
Demand Management
Advanced Forecasting and Demand Modeling
Real-time Sales & Operations Planning
$900K license and $300K for implementation
35. The Planning Dashboard
Direct access to
online reporting and Key Performance
personalized Indicators
worksheets showing current
Quick Data Access status of
important
information
Focus planners
attention
Automated workflow and
Exception Management of
business processes reduces
information handling
Handling service levels
before it is a problem
36. True Demand Logic
Advanced Logic to „clean‟ demand prior to forecasting
Sell Outs
Estimated Returns
Estimated and Projected Subscriptions
Hidden Demand
37. Safety Stock Calculation and Review
Availability
calculated based
on Service level &
Demand
variability
Agent Band re-
calculated each
week,
Demographics
loaded to allow
focus on key areas
39. Return on Investment
Budgeted savings exceeded for all 7 days for a 6 month period
Change in returned copies: July-December, 2005
Budgeted Actual
Weekday -8% -11%
Saturday -1% -5%
Sunday -2% -7%
• Reduction in returns of Sydney Morning Herald by 15% and 5-
10% for other publications
• Increase in availability levels
• Expected savings of $300K/month and have exceeding this
41. Case Study – National Brands Ltd
Supply Chain: Marketing: Sales Marketing:
Build stock for planned Sales: Decide to check with Marketing if Confirms there is
maintenance shutdown run a major promo. No there is sufficient sufficient stock. No one
review of stock levels stock,…Two days tells Supply Chain
before promotion starts about the promotion
Supply Chain: Customer: Very CEO: Very Result: Everyone
unhappy. Receives call unhappy (including
Alert…Stock levels unhappy. Not getting
from irate customer consumer) , both
dropping fast due spike stock. Consumers not
advising he will sue for Manufacturer and
in sales happy either
lost sales Retailer lose sales
Diagnosis: Poor Internal Collaboration, Poor Forecasting, Poor Promotion Management
Business Impact: Costs Increased, Profit Decreased, Customer Service Decreased.
42. La Collaborazione produce sempre i migliori risultati
Logistics
Ship
Finance
Make, Buy, Plan
Revenue and Cost
Contol
Supply
ERP
Ordini e Previsioni
Demand Planner
Sales
Demand
Strategie di
Crescita
Marketing
New Products
Think Tank
43. Sales & Operations Planning
La Collaborazione produce sempre i migliori risultati
Logistics
Ship
Finance
Make, Buy, Plan
Revenue and Cost
Contol
Supply
ERP
Ordini e Previsioni
Demand Planner
Sales
Demand
Strategie di
Crescita
Marketing
New Products
Think Tank
45. Sales & Operations Planning
Sales Budget Quantity (Manual + Statistical)
Value calculated by List price
Forecast Assumptions, Service Level, Inventory
Stocks, Demand Variability MAPE
Simulate your best Supply Plan based on Demand
and Make/Buy Constraints
Bring Financial Constraints to the table (Revenues or
Costs)
Approve the Final Demand, Supply & Finance Plan.
Execute and Monitor
48. Applica and Demantra
Applica
• $600 million + provider of consumer durables
• Principle businesses include small household appliances and
professional hair care products
• Trade names: Black & Decker Home, Littermaid and Gold-N-Hot
• Distribute across the Americas to all retailing channels
• Recently acquired by Harbinger Capital Group – Private Equity
Demantra
• Live with Demantra 6.2 in August, 2004
• Demantra is a Tier 1 system supporting Demand Planning
• Forecasted accounts ~ 65
• Active forecasted items ~ 2000
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.”
49. Applica Historical Forecast Bias
22.0%
17.0%
12.0%
Bias
7.0%
2.0%
Targeted Zone
-3.0%
-8.0%
Sep-05 Oct-05 Nov-05 Dec-05 Jan-06 Feb-06 Mar-06 Apr-06 May-06 Jun-06 Jul-06 Aug-06 Sep-06 Oct-06 Nov-06
Month
6 Month Rolling Bias Upper Limit Target
Lower Target Linear (6 Month Rolling Bias)
Forecast Accuracy measured by Weighted MAPE
Reduction in Forecast Bias has yielded ~$9 million (13%) reduction in average
inventory.
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.”
50. Applica Historical Forecast Accuracy
85%
80%
75%
Accuracy
70%
65%
60%
55%
50% FC Acc 6 Month MA
6
05
5
06
06
6
5
5
6
5
5
6
6
6
6
6
6
-0
l-0
l-0
-0
-0
-0
-0
-0
0
r-0
-0
-0
-0
-0
n-
n-
n-
b-
ay
g
p
g
p
ov
ov
ct
ct
ar
Ju
Ju
Ap
Ju
Ja
Ju
Au
Se
Fe
Au
Se
O
O
M
M
N
N
Month
Lead time of more from about 58% in earlyweekly enterprise planning,
FC Accuracy has moved than 100 days, 2005 to near 75% in Nov. 2006
forecast accuracy improved to 80% levels
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.”
51. Processes and Systems:
• Weekly Corporate Real-Time S&OP
• Organization-wide commitment to ONE Forecast
• Demantra is the support tool and sole source for plan data
• No second guessing by Finance or Planning
• Demand Planning is integrated into weekly RT S&OP
The Results:
• Improved inventory turns from <2 in 2004 to 5 in 2006
• Total inventory reduction of ~ 33%
• Fill Rate change from 80% to 93%
• Includes virtually all 2nd tier accounts with Fill Rate > 88%
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.”
52. Case Study – Johnson & Johnson
Presentation given at Oracle
OpenWorld San Fransisco Oct 2006
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 52
53. J&J and LifeScan
• Johnson & Johnson
• World's most comprehensive manufacturer of health care products
and provider of health care services for the consumer,
pharmaceutical, and medical devices and diagnostics markets
• More than 200 operating companies under its management
• LifeScan Inc. – an operating company of J&J
• Leading maker of blood glucose monitoring systems for home and
hospital use
• Dedicated to improving the quality of life for people with diabetes
with OneTouch® Brand Products
OneTouch® Ultra® OneTouch® Ultra®2
OneTouch® UltraSmart ®
OneTouch® UltraMini™
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 53
54. Forecasting (Demand Planning) Is
Collaborative
Finance
Sales
Market Research
Marketing
Forecast Manufacturing
Supply Planning
Clinical
R&D
Customers
Multiple People = Multiple Opinions
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 54
55. Effective Demand Planning Combines Qualitative and
Quantitative Analysis to Provide Meaningful Outputs
Quantified effects
Statistical analysis
Modeling Tools
Model results
Historical Sales
Physician
Marketing perceptions
Trend message
analysis
Channel
dynamics Formulary
Customer status
Competition behavior
Supply
Price Seasonality constraints
Managed
care Promotions health care
reform
Judgment + Experience
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 55
56. Forecasting comes with its own brand of
politics
• No single person or input is able to capture the entire
picture
• Each input has its own purpose and bias
Finance Sales Marketing Supply Planning
Are we making our Am I getting Is my brand healthy? Are we meeting
numbers? compensated? Am I getting enough customer demand?
Drivers: Bottom line Drivers: Quota product? Drivers: Order
Drivers: Brand image, Fulfillment metrics,
Sales Inventory costs,
Backorder
• In the end there is no accountability
• Accuracy cannot be easily measured
• What number should supply and operations plan to?
• What number should management report to HQ?
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 56
57. The Forecast Consensus Process
Inputs Outputs
Sales (field intelligence -> short term
forecast)
Regional and franchise
Marketing (market intelligence -> consensus demand plan
long term forecast)
Forecast Error (MAPE)
Stat forecast (customer order history
-> short and long term forecast) Forecasting Forecast Changes
Finance (commitments to corporate -
> business plan)
Category/Competitive Insight
(competitive intelligence; share goals;
market data - > forecast 3 to 4 years
out)
Other tools
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 57
58. The demand planning mantra
– “What’s not in the system, does not exist”
• The „Final Consensus Forecast‟ series is the final
answer
• Numbers entered in the system get locked after end
of cycle
• SKU level information is rolled over to Supply and
Operations group for planning purposes
• Numbers in the system are used for all S & OP,
Sales, Marketing and Finance related discussions
• Forecast reports are generated off of numbers in the
system
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 58
59. S&OP Process Overview – S&OP
activities, inputs and outputs
Supply/Demand balancing and scenario planning with a medium to long term focus
Global Forecast
Identify projected Directions (to DP)
supply/ demand
Franchise imbalances
Consensus
Forecast Review and
(from DP) discuss future Senior Mgmt.
outlook and review and
scenarios with Preliminary Global
Supply/Demand
approval in
Execute key supply & Balancing Executive
Portfolio demand Recommendations S&OP
Review stakeholders in Meetings*
S&OP Meetings
Supply Information: Supplier forecast
Capacity, Lead Execute Supply (via SP to suppliers)
Times, etc. Review
Execute
Financial
Review
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 59
60. The Flip Side of a Single Number
Consensus Process
• Occasional need for offline communication:
• Upsides/Downsides to forecast (Market intelligence)
• What-if scenarios and contingency planning
• Major changes since last forecast lock
• Sometimes there is really no single number:
• Competitive product launches - Large uncertainty in outcome
• Internal new product launches - Large range in forecast
• The politics of single numbers:
• Internal new product launches – Different groups have different
opinions on launch dates
• Numbers in the system may not meet needs of all the groups
involved
• For example, production may use MAPEs/experience/inventory
policies on top of the number in the system for planning
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 60
61. The Key To Success With a Consensus
Forecasting Process
• A well developed demand forecast process combined with a
strong S & OP process
• Process compliance
• Process is more important than the number itself
• Unbiased group that acts as a liaison between all involved
groups
• Assumption based forecasting
• A forecast is as good as its assumptions
• Visibility to assumptions drives belief in the forecast
• One voice
• Effective communication
• Understanding that range is NOT a bad thing, but having multiple
numbers floating around IS
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 61
62. Advantages of a Single Number
Consensus Forecasting Process
• Everyone speaks the same language – ONE VISION
• Range and uncertainty is still correctly captured and
communicated in forecast assumptions and upsides/downsides
discussions with planners
• Marketing strategies are focused – can be measured
• Company resources are optimized
• Long Term Planning becomes easier
• Drives accountability/responsibility
• Easily measure performance – MAPE/Forecast change
• In the end – everyone knows the health of the company (Visibility
Visibility Visibility)
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.” 62
63. Oracle Demantra
Evolvere gradualmente verso la soluzione best in class
Forecast basato su attributi e
Si parte cartatteristiche prodotto
da un punto
Calcolare il lift delle promozioni e
qualsiasi l’analisi degli impatti promozionali
sulla domanda
Calcolo del forecast in base a
simulazioni di eventi
Introdurre il forecast di nuovi
Introdurre il forecast di nuovi prodotto
prodotto
Collaborare con I clienti
Collaborare con I clienti
Usare statistiche avanzate con Usare statistiche avanzate con
fattori causali fattori causali
Allet complessi con fogli di Allet complessi con fogli di
lavoro customizzati lavoro customizzati
Rolling forecasts Rolling forecasts Rolling forecasts
Eliminare
Fogli excel Collaborare per creare un Collaborare per creare un Collaborare per creare un
numero univoco numero univoco numero univoco
Usare statistiche, allert,
Usare statistiche, allert, Usare statistiche, allert,
ridurre i fogli di lavoro
ridurre i fogli di lavoro ridurre i fogli di lavoro
Creare fogli di lavoro ad hoc
Creare fogli di lavoro ad hoc Creare fogli di lavoro ad hoc
per ogni figura
per ogni figura per ogni figura
Da minor complessità a best in class
“This presentation is for informational purposes only and may not be incorporated into a contract or agreement.”