SAPSCM/APO European Initiative
APO Overview Internal Training Training Demand Planning Overview
March 2003
Training Agenda
Advanced Planner & Optimizer Overview Demand Planning Overview Supply Network Planning Overview Production Planning & Detailed Scheduling Overview Global Available-to-Promise Overview APO Integration & CIF Overview APO Implementation Considerations
Objectives Main Goals of This Section
To understand Demand Planning as an accurate forecasting tool in the APO context.
To know Demand Planning main features, in concrete:
Its architecture, data storage and representation attributes
Its main tools (Planning Toolbox, Planning Environment, Accuracy Analysis…)
The different forecasting methods available
To visualize how DP applies to a real case (Sara Lee).
To be aware of main considerations and complexity factors when implementing Demand Planning.
To get familiar with the look of DP and its basic functions through a demo and practising with simple exercises.
Contents
1.
Dema De mand nd Pla Plann nnin ing g Feat Featur ures es & Cap Capab abil ilit itie ies s
2.
Case Study: Sara Lee
3.
Key Ke y Asp Aspec ects ts to Con Consi side derr Whe When n Imp Imple leme ment ntin ing g DP DP
4.
DP Dem Demo: o: Acc Accel eler erat ated ed Sup Suppl ply y Chai Chain n Inte Integr grat atio ion n APO APO Template
5.
DP Exercises
Demand Planning Features and Capabilities Demand Planning – Accurate Forecasting
A toolkit of statistical forecasting techniques
Tightly linked to the R/3 System and the SAP BW SAP (data can be automatically transferred)
Tree selection and drill-down capabilities facilitates navigation through multidimensional multidimensional data structures
Uses the Alert Monitor to report exceptions Planner’s Knowledge Task-specific planning tools
Information
Collaborative forecasts
Order & shipment actuals & history
Cost
POS data
Nielsen / IRI data
...
Flexible views Graphics Promotional planning Life cycle management Cannibalization Accuracy reporting
Demand Planning Data Mart
Statistical Methods Multi-model approach
Average models Exponential smoothing Causal factors Trend dampening Model combination
Anticipation of Future Demand
Demand Planning Features and Capabilities APO Demand Planning within Supply Chain Planning
BUY
MAKE
DESIGN New Product Development
YEARS
MOVE
HOLD
SELL
Logistics Network Design
Demand Sales Planning Forecasting APO – DP
QUARTERS
S
Supply Contract
Network
Customer Service
Negotiations
Sourcing
Territory Planning
U Production Planning
P MONTHS
P
Material Requirements
WEEKS
Planning
I E Procurement
DAYS
R HOURS
Supply Demand Matching
Materials Planning
L
Inventory Target Setting
Material Inventory Tracking
B2B Exchanges
Detailed Production Scheduling Manufacturing Execution System Production Activity Control
Contract Manufacturers
Product Allocation
Transport Planning
Load Planning
C U S T O M
Distribution Requirements Planning
Available to Promise
R
In-transit & On-hand Inventory Tracking
3PLs / 4PLs
E
Order Management
Channel Partners B2B Exchanges
Demand Planning Features and Capabilities SAP APO Demand Planning Architecture
Demand Planning is composed of three layers:
Graphical user interface
Planning and analysis engine
Data mart
Planning Views
GUI
Planning & Analysis Engine
Data Mart
OLAP Processor
Business Planning Library
Statistical Forecasting Toolbox
Planning Area
Time Series Catalog
Notes
Demand Planning Features and Capabilities SAP APO Demand Planning Architecture (continued)
Performance is of vital importance in any demand planning solution if users are to fully benefit from available information
DP architecture includes several features to ensure high performance:
Dedicated server
Multidimensional Multidimensional data mart based on the star schema that supports efficient use of storage space and of CPU cycles, minimizing query response time
Batch forecasting so do not impede online performance
The size of the information treated depends on:
Number of characteristics: characteristics: many characteristics will let the user more flexibility to define the planning level and to review the information but it makes the system works slower
Number of key figures: figures: many key figures will give the user a lot of information related to forecast but it makes the system works slower
Number of characteristic combinations: combinations: the time consuming for any calculation (e.g. macros) depends directly on the number of characteristic combinations
Number of planning versions: versions: two planning versions needs double capacity than one
Type and number of temporal periods
Demand Planning Features and Capabilities Data Storage and Representation
Multidimensional Multidimensional Data Storage in the data mart allows to:
View data and plan from many different perspectives
Drill down from one level to the next
Info Cubes: Cubes:
A multidimensional multidimensional data structure
The primary container of data used in planning, analysis and reporting
Contains two types of data, key figures and characteristics (or dimensions): - Key figures are quantifiable values (e.g. sales in units, orders, shipments, POS…) - Characteristics or dimensions determine the organizational levels at which you do aggregation and reporting (e.g. products and customers)
Info Cubes also share master data and descriptive text, which are stored in different tables
The Online Analytical Processing processor:
Models the business rules considering the aggregational behavior of key figures (e.g. sales summed by product and time)
Guarantees that all business rules are met and the computed views present valid results
Demand Planning Features and Capabilities Data Storage and Representation (continued) (continued)
Hierarchies are modeled as combinations of characteristic values (e.g. product are grouped into product family hierarchies) using proportional and temporal factors, in order to be used as the basis for aggregation, disaggregation and drilling down.
The DP planning level is based on the characteristics definition. In order to be more integrated with R/3 data, the dimensions and characteristics are usually based on R/3 hierarchies:
Product dimension and characteristics are usually based on R/3 product hierarchy
Customer dimension Customer dimension and characteristics are usually based on a R/3 customer hierarchy
Geographic dimension and characteristics are usually based on the supply network
Dimensions
Facts Aug.
Sept.
W 3 2 W 3 3 W 3 4 W 3 5 W 3 6 W 3 7W 7 W 3 8 W 3 9 W40 W41
Hierarchies Regions
i o e r t P s e d 203 u m r C o 124
Material
Attributes
Life Cycle
time sequence
Product Groups
Promotion
Forecast
Time Series Management ‘99
‘00
‘01
‘02
N o t e s
Demand Planning Features and Capabilities Data Storage and Representation (continued) (continued)
Time Series Management: Management :
Based on catalogs: time series data with related attributes (e.g. promotional patterns and life cycles)
SAP DP allows to reuse time series saving time and ensuring consistency (e.g. reuse a past promotional pattern to estimate the impact of a similar future promotion)
Notes Management maintains all notes entered by planners to create an audit trail of all demand planning activities, which is specially helpful when multiple sources and people are involved (such as in consensus forecasting)
Dimensions
Facts Aug.
Sept.
W 3 2 W 3 3 W 3 4 W 3 5 W 3 6 W 3 7W 7 W 3 8 W 3 9 W40 W41
Hierarchies Regions
i o e r t P s e d 203 u m r C o 124
Material
Attributes
Life Cycle
time sequence
Product Groups
Promotion
Forecast
Time Series Management ‘99
‘00
‘01
‘02
N o t e s
Demand Planning Features and Capabilities Planning Environment
DP’s rich planning and forecasting functions are based on the Statistical Forecasting Toolbox and the Business Planning Library. Library. These functions include:
Aggregate functions (sum, weighted sum, sum, average)
Disaggregate functions (quotas, proportional and equal distribution)
Comparison functions (difference, ratio, percent, percent difference, share and correlation)
Financial functions (conversion from units into revenue, currency conversion and business period conversion)
Time-series functions (time-phased, average, and weighted average of time series)
A Planning Book is an easy-to-use tree control for selecting data and a frame with a grid and a graphical data display: Preconfigured
planning books for promotional planning, causal analysis, statistical forecasting, life cycle management, etc These
can be used as guides for customized planning books
Demand Planning Features and Capabilities Planning Environment (continued)
You can use Advanced Macros to:
Calculate deviations
Make automatic corrections
Calculate sales budgets
Define your own exceptional situations
Launch status queries
Advanced Macros models the calculations calculations based on the individual individual business tasks to perform principally:
Build a macro consisting of one or more steps
Control how macro steps are processed and how results are calculated
Use a wide range of functions and operations
Define offsets so that the result in one period is determined by a value in the previous period
Restrict the execution of a macro to a specific period or periods
Write macro results to a row, a column or a cell
Create context-specific and user-specific planning views
Trigger an alert in the Alert Monitor to inform of particular business situations
Integration with Microsoft Excel
Demand Planning Features and Capabilities Statistical Forecasting Toolbox
A Toolbox of all practical, proven forecasting forecasting methods Time Series Models:
Uses past sales to identify level, trend, and seasonal patterns as a basis for creating future projections
Naïve models, moving average, simple linear regression, Brown’s exponential smoothing, HoltWinters, Box-Jenkins
Stochastic Models:
Accurate forecast with sporadic demand pattern pattern Croston model uses exponential smoothing to estimate: - The size of demand during periods in which demand occur - The demand frequency
Final forecast are determined by distributing the size of demand according to the demand frequency
Forecasting
Demand Planning Features and Capabilities Statistical Forecasting Toolbox (continued)
Multiple Linear Regression:
Technique for estimating the relationship between past sales and other causal factors
Variety of options to model linear and non-linear trends: - Seasonal patterns - Life cycle patterns - Dummy variables and time lags
Correlation analysis corrects variables
Pick-the-Best, applies the best method among:
All of the available forecasting methods, methods, or
The planner-specified forecasting methods
S-Shaped Curves supports complete lifecycle forecasting (introduction/growth and end-of-life phases)
Logistic and exponential functions
First estimation based on similar products
Adjusted over time when sales sales history is available
Demand Planning Features and Capabilities Causal Analysis
Includes all significant causal factors (price, number of displays, number of stores, temperature, working days…) in the models and determine how they affect customers’ behavior
Simulate sales development according to the mix of causal factors (what-if analysis, marketing mix planning)
Multiple linear regression to model the impact of causal factors
50°F
65°
75°
85°
60°
s e l a S t i n U
Feb.
Mar. Ma
April
May Ma
June
July
Aug.
Sept.
Demand Planning Features and Capabilities Multi-Tier Forecasting
Integrates sell-in data (like POS data) into the process of forecasting sell-through data (like shipments)
Causal model based on significant causal factors to forecast POS
Second causal model is used to forecast shipments:
Uses past POS data and the POS forecast as the main causal factor
Takes the time lag between POS and shipments into account
Considers other causal factors (forward buys, trade promotions…)
Manufacturing
Sales History
Sell Through Sell In
sales n n o o i i t t o o m m o o r r P P r o t i t e p m o C
t
n e m e s i t r e v d A
Retailer
POS Data
Consumer
Consumer demand + Replenishment lead time + Forward buying = Retailer Demand
time
Demand Planning Features and Capabilities Data Analysis
Identifies missing values and outliers in the data to improve the quality of the statistical forecast. Through the outlier, an automatic correction of historical data is done taking into consideration out-ofrange data that may disturb the identification of historical pattern
Identifies structural changes in “established” patterns:
Level, trend, and amplitude changes
Change from unstable to stable behavior
Automatic detection via tracking signals signals
Automatic outlier detection & correction correction Manual intervention
Demand Planning Features and Capabilities Promotion Planning
Impact of promotions must be projected separately from standard forecast components that are based on historical sales data
Takes prices into account when doing profitability analysis for promotional calendars
Reporting capabilities allow to track promotional activities and related costs
Archives a promotion pattern in a promotion promotion catalog, so it can be reused Several techniques for estimating the effect of a promotion Forecast simulation Sales
Profit Promotion Promotion patterns
-10%
‘97
‘98
‘99
‘00
Price
Planner Quantity
Demand Planning Features and Capabilities Life Cycle-Managem Cycle-Management ent
A Demand Planning Planning and Supply Network Network Planning both components’ components’ function function Planning strategies for a product depend on the stage of its life cycle :
Should the product be introduced, and when?
How should a product be promoted during the different stages?
Should the product be deleted, and when?
Should a successor product be introduced?
Should a re-launch be started for a product, and when?
What is the cannibalization effect of a new product with existing products? Etc.
DP can represent the launch, growth and discontinuation phases by using phase-in, phase-out and like modeling profiles (or combining them):
A phase-in profile reduces demand history by ever increasing percentages percentages during a specific period or periods (simulating upward sales curve – launch and growth phases)
A phase-out profile reduces demand forecast of a product by ever decreasing percentages (simulating downward sales curve – discontinuation discontinuation phase)
Like modeling creates a forecast using the historical data on a product with a similar demand behavior (new products and products with short life cycles) Product Launch
End of Life
Aggregate
Demand Planning Features and Capabilities Consensus-Based Consensus-Ba sed Forecasting
SAP DP supports consensus-based Sales & Operations Planning (S&OP)
Multidimensional Multidimensional data structure of the InfoCubes I nfoCubes enables to createmultiple create multiple plans: plans:
Product levels for Marketing
Sales areas and account/channel for Sales
Distribution centers and plants for Operations
Business units for Finance
Synchronizes multiple plans into one Consensus Plan that drives business
Composite Forecasting reconciles and combines different plans on same level and multi-levels Sales Forecast 1
Forecast
... Marketing Forecast n
Combine & Reconcile
Demand Planning Features and Capabilities Forecast Accuracy Analysis & Alert Monitor
Forecast accuracy reporting: reporting:
Helps to assess the accuracy of past forecasts
Integrates this knowledge into projections for the future
Stores a series of forecasts for a particular period and compares each deviation of this series to the actual values for the same period (mean absolute deviation, error total, mean percentage error, …)
Reports shoe forecast errors at any level and dimension:
Actual versus forecast
Actual versus time-lagged forecast
Actual versus different planning versions versions
Actual versus budget
Alert Monitor informs Monitor informs in real time via e-mail or exception message if an exception occurs
Exception conditions can be defined based on thresholds for special statistics and tracking signals
Reports can be sorted:
By forecast error
Restrict them to products with a forecast error greater than a specified threshold
Demand Planning Features and Capabilities Advantages of SAP APO Demand Planning
Global server with a BW infrastructure
Integrated exception handling, creation of user defined alerts
Integration with Production Planning (S&OP scenario)
Main memory based planning
Flexible navigation in the planning table, variable drill down
Extensive forecasting technique
Promotion planning and evaluation
Collaborative planning via the internet
Supports Sales Bills of Material (BOMs)
Contents
1.
Dema De mand nd Pla Plann nnin ing g Feat Featur ures es & Cap Capab abil ilit itie ies s
2.
Case Study: Sara Lee
3.
Key Ke y Asp Aspec ects ts to Con Consi side derr Whe When n Imp Imple leme ment ntin ing g DP DP
4.
DP Dem Demo: o: Acc Accel eler erat ated ed Sup Suppl ply y Chai Chain n Inte Integr grat atio ion n APO APO Template
5.
DP Exercises
Case Study: Sara Lee Introduction
Main objectives of Demand Planning for Sara Lee:
S&OP purposes: purposes: Provide the essential input for S&OP monthly cycle (forecast) and create consensus within the OpCo. Demand Forecast should contain the required detail in order to compare with Business/Sales targets
Supply Planning purposes: purposes: Provide updated forecast from different OpCo’s (in weekly buckets) to Supply Planning in order to base Supply Planning on consolidated consolidated forecast from each OpCo
Benefits of Demand Planning for Sara Lee:
Improve the communication and transparency from all OpCo’s to CoE
Provide to Supply Planning short and long term volume estimation for capacity planning
Create consensus in the OpCo (together (t ogether with S&OP)
Understanding the demand of each OpCo through t hrough deep analysis (KPIs, market intelligence,…)
Move from ”Reaction on” toward ”Plan Activities”
Improved customer service level
Lower obsolete and safety stocks
Case Study: Sara Lee Project Approach
A template has been developed in order to align, cover and support all the processes performed in the Sara Lee Opcos in Europe. In different phases, the Opcos will start to use the new template, changing their actual procedures and/or systems (local roll-outs).
There will be a central team responsible of maintaining the basic and common applications.
In every roll-out a local team will be assigned to check that the requirements of the Opco are covered, to conduct the trainings, etc.
Communication Communication between local and central teams: teams:
Either in the central and in the local teams, there will be a member responsible of the communication between them. The communication link will be one-to-one.
CUSTOMIZING: CUSTOMIZING: The local team will ask the central for customizing new structures.
Every local roll-out will have a different copy of the “Implementation Guidelines”.
GAPS: The local team will detect functionality not covered by the template, then, these gaps must be written down in a document called “EuRoPe fit”.
Both teams will have a meeting to determine how each issue in the “EuRoPe fit” must be solved.
Case Study: Sara Lee Project Approach (continued)
Procedure for the “EuRoPe fit” Analysis and Development:
Initial training (central to local)
EuRoPe fit sessions (local)
EuRoPe fit analysis (central & local)
GAP estimation (central)
GAPs approval (project management)
Local GAPs design (local)
Central GAPs design - Template development (central)
Case Study: Sara Lee Demand Planning Processes
Demand Planning Processes are divided into three cycles: cycles:
AOP/Outlook AOP/Outlook generation: generation: Provide volumes taken from APO DP as a starting point for the AOP/Outlook generation
Monthly cycle: cycle: Update Demand Forecast for the following 24 fiscal periods and provide it to the Sales and Operations Planning monthly cycle (to create a consensus and run Supply Planning).
Weekly cycle: cycle: Review current month forecast to identify supply risks, advise Sales and Marketing of these risks and change the forecast which applies to a period outside of the Supply Planning frozen period.
Strategic
Tactical
Operational
AOP generation
Monthly Cycle
Weekly Cycle
Case Study: Sara Lee Demand Planning Processes (continued)
Process
AOP/Outlook AOP/Outlook generation: generation:
APO
APO Forecast volume can be be used as a starting point for AOP AOP generation. generation. Volumes are sent to R/3 where it is converted into value.
CO-PA (R/3)
Volumes from APO DP
Responsible Demand Planning
Convert volume to value Finance
Adjust Volume
Volume/value adjustments are done in R/3 AOP volume is sent back to APO APO for Supply Planning purposes and KPI analysis
Send Adjusted Volume to APO APO
Run SNP with Adjusted volume Volume adjusted after SNP
CO-PA (R/3)
Supply Planning
Convert volume to value Finance
Adjust Volume Interface SAP - APO
Final AOP volume sent to APO
