1. APO
APO is the APS (Advanced Planning and Scheduling) solution from SAP. This provides for more advanced features than standard ERP like R/3, and incorporates various business partners (i.e., vendors, customers,) in the planning aspects. APO is a pure planning tool (strategic and tactical level), the posting of the transactional data will always reside in R/3 system.
Record transactional data, then update BW
1. The planning takes place within SAP-APO. Demand Planning pulls the data to be used from the R/3 system.
2. After the plan/forecast analysis has taken place, the plan is pushed back into R/3 for actual execution.
3. The Data Warehouse is used to store that transactional data which will be used for planning and analysis purposes.
Range of Planning Functions
• The diagram below is to illustrate how the various planning horizons can vary within APO and their applications. For example, Demand Planning starts with the longest view into the future. As the plan becomes more refined, based the area, the horizon becomes smaller.
2. Features of APO-DP
DP is the APO module dealing with demand. It includes forecasting (statistical forecasting based on sales history and causal factors), market intelligence (corrections and promotions), product life cycle management, consensus planning and collaborative planning.
The interactive planning book groups together various Demand Plans.
We can maintain various demand plans (statistical forecast, correction, marketing plan, promotion plan, customer plan).
3 4
2 5 1
Planning book features:
- Navigation: is very user-friendly. Most transactions can be accessed from this screen (access to forecasting, promotion, macro definition, planning book/data view definition).
- Aggregation/Disaggregation: APO-DP’s method of aggregation and disaggregation is referred to as a “consistent” planning model. We can enter forecasts at any level of the hierarchy, values at higher levels are calculated automatically by summing up, and values at lower levels are calculated using proportional factors. APO provides several ways to calculate automatically those proportional factors.
- Fixing values: We can fix cells so that the forecast values are not changed automatically by the system. A small icon inside the cell alerts the user that the value has been fixed.
- Entering notes: we can use notes to explain (for our own information or for any other person accessing the data) the reasons why a sales forecast, for example, is particularly low or high at a particular period and on a particular level. A small icon inside the cell alerts the user that a note has been entered.
APO is very user-friendly, using drag and drop, drill-down, slice and dice. The interactive planning book looks like an Excel spreadsheet. Any result can be downloaded to Excel.
We can plan promotions or other special events separately from the rest of our forecasts. We can use them for one-time events or repeated events such as quarterly advertising campaigns. We can keep track of the different types of promotions: trade fairs, coupons, freestanding inserts, competitors’ activities, market intelligence.
- We can compare forecasts with promotions to those without promotions
- We can correct the sales history by extracting past promotions from it, thus obtaining unpromoted historical data for the baseline forecast
We can use life cycle planning to forecast the phase-ins of new products and phase-outs of old products. With “like” modeling we can create forecasts for a new product using the historical data of a product with similar sales behavior.
These models investigate the historical data from constant, trend and seasonal patterns, and then issues relevant forecast errors. This is the most widely used type of models because of their simplicity.
This is the only type of forecasting in R/3.
- APO-DP calculates concurrently 6 forecast errors: MAD (Mean Absolute Deviation), ET (Error Total), MAPE (Mean Absolute Percentage Error), MSE (Mean Square Error), RMSE (Root of the Mean Square Error) and MPE (Mean Percentage Error).
- Automatic outlier correction: If a historical data lies outside of a defined “tolerance lane”, it will be corrected based on parameters defined in the forecast profile.
- Workdays correction: forecasts will be adjusted based on the number of workdays maintained in the factory calendar
- Forecast versions: They are used to save and compare the parameters from several forecast runs. Both forecast errors, and model used (including parameters, and users who have already forecasted this selection) are saved. After several forecasts (with different models and parameters) are executed, we can sort by forecast error to find out the best model and the best parameters.
- Automatic model determination: Historical values can be tested for trend, season or trend/season. The system then chooses the model which displays the lowest MAD.
MLR is used to correlate a dependent variable (sales, for example), with independent variables (such as prices, advertising, and seasonal factors).
Causal analysis is based on causal factors such as prices, budgets and promotions. The system uses MLR to calculate the influence causal factors had on the sales history, and thus allows an analysis to be made of the success of specific actions. The connection between causal factors and sales history that was calculated is used as a basis for modeling future actions.
- Composite forecasting: We can weight any number of models and combine them to get a final statistical forecast.
A consensus-based sales and operations planning process is one in which we view forecasts from various departments/customers and integrate them into one-number consensus forecast. This then drives the business.
This allows customers or partners for example, an input via the Web to communicate forecasts.
This informs about exceptional circumstances that have originated in the plan. DP alerts are generated through macros and through the forecast. As the macros can be user-defined, we can have any type of custom alerts.
Live transaction data in APO are in the active version 000. This version can be copied and used as a basis to simulate different scenarios.
Advanced macros are user-defined formulas that can be executed within the grid of a planning book. They are pretty much like spreadsheet formulas, but provide more flexibility and function better than typical formulas in spreadsheets.
All operations, mathematical functions and special functions such as “number of working days” in the period are available to the users.
There is no programming language involved, the building of the macro is done through drag and drop technique.
APO has its own application server. Transactional data is still entered in R/3 while the heavy processes such as forecasting and analysis are done in APO, thus keeping a good response time in the R/3 system.
Comparison between R/3 and APO
FEATURE R/3 APO
Interactive Planning Book Very poor Powerful
Graphical User Interface Very poor User-friendly
Univariate forecast models. 6 errors are available. Exists. Error MAD only available More models, more error calculation, comparison
Multiple linear regression and Causal analysis No Yes
Consensus forecasting No Yes
Collaborative planning No Yes
Promotion management No Yes
Life cycle management and Like modeling No Yes
Alert monitoring No Yes
Simulation No Yes
Macro Exists but very difficult to build. Yes. Very flexible, no programming language required
Separate application server No. Impact on response time. Yes. Response time remains good.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.