SAP IBP for Demand

Objectives
After completing this lesson, you will be able to:

After completing this lesson, you will be able to:

  • explain What's New for Demand

SAP IBP Demand Forecasting 2302

Minimum values for AutoARIMA parameters ranges

Fine-tuned exhaustive model search

Control option for forecast job status determination

Set failure / success criteria

Forecast Automation enhancement

Save residual and trend components into key figures

Unified authorization handling in forecast jobs

Reconciled user experience in Excel and Web UI 

Manage Product Lifecycle - Overall Process and 2305 Enhancements

Scenario:

  • From time to time a company has to introduce new products in the current product portfolio. This can be caused by either legal reasons, by the need of new packaging or the new product can establish an absolutely new product category for the company.
  • Demand planning aims to create a high quality automated forecast for future demand of a product. This is done be sophisticated statistical algorithms that calculate a future time series with the knowledge of past values. For new products, such past values, which equals the number of sold items of the product in the past, do not exist. This problem can be solved by manual creation of forecast data or by copying historical data of reference products to the new product.
  • The best way to deal with new products is nevertheless to define reference products for the new product that are taken into account dynamically in the forecast engine at the time of the forecast run.
  • Improve Forecast Accuracy for new products
  • Reflect the common phase-in or phase-out for new or obsolete products and generate a reliable plan for products w/o history

  • Cover Like Modelling with many reference products, weights and offsets/validity dates
  • Phase-In and Phase-out with user defined curves
  • Supports scenarios of product replacements and introduction of new product lines.
  • Currently used in statistical forecasting and demand sensing calculations

The Product Lifecycle Process is one of the key topics in the Demand Planning area.

This is the case because the introduction of new products to the market as well the the elimination of obsolete products from the market are critical times in the lifecycle of a product that needs special attention from a planners perspective.

Especially the introduction part is a fragile process with losts of potential to loose sales. Example: Marketing Campaign started, customers want to buy the new product, but it is already out of stock in the supermarkets because of wrong planning.

  • Like Modelling: leverage the sales history of an existing reference product to forecast a new product.
  • Phase-In: Leverage different phase-in curves to introduce the product. Usually, sales slowly picks up or shows an exponential trend and then flattens after some time. We have already made good progress in the last releases where we added the Like Modelling piece as well as the support for Phase In. In IBP, we already have some pre-defined phase-in curves and are now moving towards user-defined curves.
  • Phase-Out: Other way round. Product is slowly taken off the market using excess inventory. Maybe it is also replaced by another product. This is on the roadmap for 1711.

There is a specific Fiori app where all the relationships and configuration can be maintained. This app provides a

  • Graphical User Support (chart on the lower right side showing a graphical representation of sales history with multiple reference products „like modelling")
  • Flexible Modelling Capabilities (Chart showing Phase-In on the upper right side)
  • High Level of Automation - especially during statistical forecasting, the settings in this app are automatically taken into account.

Currently, the Product Lifecylce Management is used during Statistical forecasting and demand sensing in the demand area. However, it is also planned that other IBP solutions like inventory management leverage these capabilities in the future (see next slide).

2 References can be defined for products (in general the level 1 attribute of the assignment level) via the Manage Product Lifecycle Fiori app.

  • The intention is to use the historical values of a reference for forecasting.
  • For product references weighting factors can be defined
  • Several references can be defined with validity dates or offsets for one product attribute value in order to construct useful historical data.
  • Assignments of references can be active or inactive. During save it is checked if planning objects exist for product and reference product. Assignments are deactivated if no planning objects exist.
  • The assignment level is defined in the Settings for Product Lifecycle app per planning area. The first level is an arbitrary product attribute and two additional non-product attributes can be chosen.

2 Forecast Dates can be defined for a product (in general the level 1 attribute of the assignment level) and values of a single launch dimension.

  • The intention is to model the shape of the phase-in and phase-out behavior of the forecast.
  • Forecast start dates can be defined. Before this date no forecasting will be executed.
  • Phase-In start and end dates and phase-in curves can be assigned. The phase-in start date marks the point of time from which the new product is sold in the market. No forecast is generated in time buckets before that date.
  • Phase-Out start and end dates and phase-out curves can be assigned. The phase-out end date marks the point of time from which the product is not sold any more. No forecast is generated in time buckets after that date.

2 Forecast Dates can be defined with parameters for manual forecasting.

  • The intention is to create a forecast for the new product based on assumptions and not based on references providing historical values for the forecast.
    • This approach can be used if no reference is available or a reference cannot be easily found or the possible references do not fit for all aspects like trend and seasonality.
  • It can be decided by product if manual forecast should be applied.
  • No product assignments are defined in this case.
  • Forecast start dates, Phase-In start and end dates are identical to standard forecast dates definitions.
  • No phase-out settings are defined here.
  • A base value has to be maintained (up to 2 decimal places), optionally a trend value and its dampening factor phi (0<phi<1) and a seasonality curve can be defined. For each product you specify in which unit of measure the values are maintained. Optionally a periodicity ( weeks etc.) can be attached.
  • The base value has to be greater than 0. For the wildcard assignment a base value of 0 is allowed with meaning that for all launch dimension values not specified standard forecast is executed.
  • The forecast is calculated starting from the forecast period containing the phase-in start date (forecast(1)) as follows
    • Without trend dampening: Forecast(i) = (base value + trend x [i-1]) x seasonal index(i)

2 Forecast is only executed for launch dimension values with manual forecasting parameters.

  • You can define a setting for the wildcard entry where base value = 0. This means that for all launch dimension values for which no explicit setting is defined a forecast is executed with the algorithm defined in the forecast model using the own history of the product.
  • Stop manual forecasting of a product by removing the flag "Manual Forecast" for the product.
  • To stop manual forecasting for a single launch dimension value you may specify a Manual Forecast End Date. If the current date is after the Manual Forecast End Date a forecast is executed using the parameters defined in the forecast model.
  • Forecast can be carried out for existing planning objects for the target key figure of forecasting. No planning objects are needed for the input key figure. Filters for the forecast job execution cannot be considered.
  • If phase-out definitions should be done for a product, the product must not be used in manual forecasting. To achieve this the product's definitions have to be deleted in the Manual Forecasting tile.
  • You are able to upload parameters for manual forecasting with the external OData service.
  • Manual Forecasting parameters are maintained in a single tile in the demand planner area of the launchpad for quicker and more focused maintenance.
    • There are business cases where you want to forecast on a higher level than Product ID, maybe because the data is less fluctuating and therefore the results promise to be more accurate or you do not want to do forecasting for many items. Nevertheless you defined phase-in and/or phase-out dates on Product ID level and you want the forecast to consider these settings. In this case you need the aggregated planning for phase-in and phase-out.
    • With a flag in the Manage Forecast Model app you can switch on this function. With manual forecasting switched on, aggregated lifecycle planning is not possible.
    • This function can consider definitions of forecast dates for a product and launch dimension values in a forecast run even if the level 1 attribute Product ID is not part of the forecast level but another attribute of the product master data type like Product Group is part of the forecast level.
    • In addition a forecast run with a forecast level that does not contain the launch dimension or the root attribute of the launch dimension can consider definitions on that detailed level if an attribute of the product master data type is present in the forecast level.
    • Aggregated planning is not supported for manual forecasting.
    • Prerequisites:
      • The launch dimension is part of the base planning level of the forecast key figure.
      • The level1 attribute (very often Product ID) is root attribute of the base planning level of the forecast key figure.
      • The level 1 attribute (very often product ID) is the root attribute of the product master data type.
  • Algorithm for Aggregated Planning for Phase-In and Phase-Out
  • Do a forecast on aggregated forecast level (e.g. on product group level)   

            

Save forecast result to forecast key figure in a planning session using disaggr.

  • LOOP OVER launch dimensions used in planning area
    • Read forecast result from planning session on phase-in/phase-out level (product id/launch dimension)
    • Apply phase-in / phase-out factors, if forecast start date in future, delete values
    • Save / update forecast result to planning session using disaggregation
  • Read forecast result from planning session on forecast level
  • Continue original logic: Save forecast result
  • Runtime of algorithm increases with the number of launch dimensions.
  • Disaggregation settings of the forecast key figure affect the result heavily. It does not make sense to use a key figure for disaggregation that contains the phase-in/out information together with the implemented algorithm. If such a key figure exists you can forecast on aggregated level without using the algorithm.
  • The aggregated value on forecast level ( product group level ) may be reduced with the algorithm. This can have the following reasons:
    • Phase-in / phase-out curves having percentages lower than 100% are applied on detailed level.
    • The result can be set to zero on detail level because of a forecast start date in the future.

3 Within the Manage Forecast Models Fiori app, the flag "Consider product lifecycle information" needs to be set for the individual forecast model if the system should consider references and forecast dates. The forecast model can be set up for statistical forecast algorithms or demand sensing.

  • If the above flag is set, the flag "Manual Forecast" can be set. In this case forecast engine is creating a manual forecast for the product and launch dimension values for which appropriate parameters are defined. For all other products a forecast is generated using the products own history with the algorithm of the forecast model.If manual forecasting is flagged no demand sensing forecast model can be selected.
  • You can set either the flag " Manual Forecast" or the flag "Aggregated Lifecycle Planning for Phase-In and Phase-Out".

4 You need planning objects for the new product before you can run forecasting. In detail you need planning objects for the key figure which represents the main input for forecasting in the historical horizon and you need planning objects for the target key figure of forecasting in the forecast horizon.

  • The forecast engine is on the fly using the combined historical data of the defined references for the given product attribute values.
  • If you want to use the own history of the new product as quickly as possible, you may assign the new product itself as a reference product with a weighting factor of 100%. If all reference products are not sold any more, the system will only use the sales data of the new product in this case after product phase-in.
  • If you want to stop the usage of references for a certain product you may delete or deactivate them via the Manage Product Lifecycle app or you use a forecast model with "Consider product lifecycle information" unchecked. Forecast engine is using the own history of a new product after the defined phase-in end date (if defined in settings app).

4 Planning objects can be created for a product in Manage Product Lifecycle app via the file upload of product assignments and directly via planning objects actions.

In addition a planning objects check is available under planning objects actions.

  • Key figures for the check are defaulted from the settings profile.
  • It can be configured if the check is searching for key figure values.
  • SAP Fiori notifications can be created if no planning objects are found and sent to many planners. By clicking on the notification you navigate in the Manage Product Lifecycle app.
  • Check results can be downloaded in a csv file. Results are available for 30 days.

4 Planning objects can be created in the baseline version by copying data of a reference in normalized planning areas

  • You select one line in the product assignments view and use Create under planning objects actions or you upload a file where only one reference for the product is defined.
  • Not more than 500000 planning objects are copied.
  • The creation via planning objects actions allows the definition of restrictions for the copy and the replacement of attribute values. A simulation shows which planning objects will be created.
  • The following characters are not allowed in the product ID of the new product: ,'"§%.
  • Planning objects can be copied for planning levels where the level 1 attribute is root attribute.
  • Planning objects can't be copied for planning levels with non-root attributes flagged mandatory in the planning area.
  • Results of the creation are written into the application log.

4 Planning objects can be created in the baseline version for manually specified values.

  • You select one line in the product assignments or the forecast dates view and use Create Planning Objects (under Planning Objects Actions).
  • In the simulation screen you select for each root and mandatory non-root attribute the values for which planning objects will be created. All selected values are multiplied out. Many planning objects can by defined this way.
  • The planning objects to be created can be downloaded into a csv file.
  • Planning objects can be created for planning levels where the level 1 attribute is not the root attribute but the respective root attribute values have to be specified.
  • Planning objects can be created for planning levels with non-root attributes flagged mandatory in the planning area and for attributes having an attribute check defined.

In order to define references for a new product in SAP Integrated Business Planning for demand, your planning area must fulfill some prerequisites. These prerequisites are checked when the user enters the Manage Product Lifecycle Fiori application.

  • Business Meaning: Define which attribute of the planning area represents the Product ID
  • The attribute with business meaning Product ID has to be of type character.
  • In order to be able to define assignments of references and to run forecasting for new products, the respective planning objects need to be created for these products first.

0 In order to define reference products or forecast dates in the Manage Product Lifecycle app for a new product in SAP Integrated Business Planning for demand, the demand planner needs certain authorizations for a specific planning area.

  • The demand planner needs the authorization to view master data of the product master data type in the baseline version in the planning area.
  • The demand planner needs the authorization to view data of key figures in the planning area.

If these prerequisites are not fulfilled the respective planning area is not available in the value helps of the Manage Product Lifecycle app.

For creation of planning objects the authorization object IBP_PLOBAV needs to be maintained in a customer role with ACTVT = '02' (Change) and IBP_MPOTYP = 'CREATE'. In addition in the object IBP_ADM you need to set ACTVT = '23' (Maintain) and ADMOBJ = 'PLAN_COMB'.

Read permission filters for products and dimension values are applied for product assign-ments and forecast dates but not for the creation of planning objects. If read permissions allow to view an assignment it can be changed, too. All filters are combined with the AND condition.

No attribute permissions are checked in the Manage Product Lifecycle app.

  • Define several references with weighting factors and offsets (defined in calendar days) or validity dates for a new product.
  • Define a launch dimension for a new product: A launch dimension describes the strategy how a new product is introduced to the market. E.g. if a company distributes the new product first in a special location (or country), the user can set "location" or "country" as the launch dimension and assign specific forecast start dates. This launch dimension is used for the definition of phase-out of a product, too.

Define forecast start dates: A forecast start date is the date when forecast execution starts. It specifies the point in time when the forecasting is switched from manual forecast creation to an automated forecast creation. A forecast start date is defined for values of the launch dimension e.g. Location = Canada has a forecast start date of September 01, 2016. This means before this date, no forecast model is run for the new product in Canada.

  • If for at least one value of the launch dimension a forecast start date is defined, all other launch dimension values are not forecasted. If there are e.g. 5 Locations and for Location 1 and 3 forecast start dates are defined, the forecast is not executed for Location2, Location4 and Location 5. If you want to avoid this behavior define dates for the wildcard (* - any attribute value) entry. The app is by default creating a wildcard assignment with a forecast start date of today.
  • If for no launch dimenstion value a forecast start date is defined, all launch dimensions are forecasted.

  • Phase-in start dates, phase-in end dates and phase-in curves are defined in the Forecast Dates area of the detail screen in the Manage Product Lifecycle App.
  • Define phase-in start dates and phase-in end dates: The phase-in start date defines the point of time where a product is sold in the market. Forecasting is not generating results for periods with an end date that is before the phase-in start date.
    • Forecast Start Date = execution start date of the forecast. From this date onwards, the new product is being included in the forecasting runs
    • Phase-In Start Date = begin of forecast horizon; and first date on which the forecasting generates data for the new product
    • Phase-In End Date = start of usage of own history. When the phase-in end date is reached in the historical horizon the forecast engine uses the history of the product itself.
  • Phase-in dates and curves are applied only to the forecast key figure in forecast horizon
  • Phase-out start dates, phase-out end dates and phase-out curves are defined in the Forecast Dates area of the detail screen in the Manage Product Lifecycle App.
  • Define phase-out start dates and phase-out end dates:
    • The Phase-Out Start Date is the point of time when the product is gradually sold less than in the maturity phase. This can e.g. happen if a successor product is already sold in the market.
    • The Phase-Out End Date is the point of time when the product is not sold any more in the market. The forecast engine will not generate results after this point of time.
  • Phase-out dates and curves are applied only to the forecast key figure in forecast horizon
  • In case a product is flagged for manual forecasting base value (obligatory), trend value, trend dampening coefficient phi and seasonality curves are maintained instead of phase-in curves and phase-out information.
  • The definition of a unit of measure for base and trend values is obligatory. By default the base unit of measure of the product is set.
  • A periodicity can be set for the product. If a periodicity is set the seasonality curve must have the same periodicity or no periodicity (= valid for any periodicity) assigned.
  • A shelf fill start date can be defined that has to be before the phase-in start date. Forecast engine does not change data between shelf fill start and phase-in start. Via Add and Edit shelf fill values can be defined, simulated and saved into a key figure.
  • Seasonality curves for manual forecasting can be maintained in the detail screen of forecast dates in the curves area or via the file upload.
  • The periods number means the number of periods per season and the values of the curve are maintained as a percentage.
  • If a periodicity is assigned the curve can only be used for products with identical periodicity. Forecast engine ignores seasonality curves or the whole product if the periodicity doesn't fit to forecasting periodicity.
  • The optional start date of a curve defines to which date the first data point belongs. With this the curve can be reused as the seasonal pattern is applied independent of phase-in start.
  • Assign preconfigured phase-in or phase-out curves: Optionally one of the preconfigured phase-in curves Linear, Superlinear or Sublinear or the preconfigured phase-out curves Linear - Phase-Out, Superlinear - Phase-Out or Sublinear - Phase-Out can be assigned for a new product for launch dimension values. These curves represent percentage time series. The percentages are multiplied with the forecast result in the phase-in/out horizon defined by phase-in/out start date and phase-in/out end date. The percentage time series are calculated during the forecasting run. The first period gets the value 0.1/0.9 and the last period is set to 0.9/0.1. The number of periods is determined by the number of time buckets between phase-in/out start date and phase-in/out end date in the periodicity defined in the forecast model.
  • The linear curve is calculated by a linear interpolation between 0.1 and 0.9. The Linear - Phase-Out curve uses the boundaries 0.9 and 0.1.
  • The superlinear curve applies the square root function to the time series that we get when we do a linear interpolation between 0.01 and 0.81. The boundaries are calculated by applying the reverse function of square root to the standard boundaries 0.1 and 0.9. The Superlinear - Phase-Out curve uses the boundaries 0.9 and 0.1.
  • The sublinear curve applies the quadratic function to the time series that we get when we do a linear interpolation between 0.316 and 0.949. The boundaries are calculated by applying the reverse function of the quadratic function to the standard boundaries 0.1 and 0.9. The Sublinear - Phase-Out curve uses the boundaries 0.9 and 0.1.

Create own function based phase-in/out curves: In addition to the preconfigured curves a user can define own phase-in/out curves in the Curves area of the Forecast Dates tab. It is possible to specify start values and end values and the function that is applied to the values interpolated between start and end values. Currently you can choose from 7 functions (Linear, Square Root, Cube Root, Fourth Root, Quadratic, Cubic, Fourth Power). A simulation showing the resulting curve is available. Parameters of the preconfigured curves can be overwritten.

  • Creating your own flexible phase-in/out curves: In addition to the preconfigured curves and function-based curves, a user can also define flexible curves. To do so, the user selects the function "single values" for any existing curve. They can then maintain values manually for each period in the table area. The start value and end value fields as well as the function field are greyed out in this case.A flexible curve must have at least 9 data points.Simulation of flexible curves is possible by changing the number of periods and choosing simulate. Before a user can run a simulation, they need to save the flexible curve.
  • Simulation of Flexible Phase-In/Out Curves
    • A Simulation shows how flexible curves are used during forecasting, if the number of phase-in/out periods differs from the defined number of periods. The result is an interpolated curve based on the flexible curve values.
    • You need a minimum of 9 data points to define a flexible curve. Simulation is possible down to 1 period.
    • The interpolation copies as many of the original values from the flexible curve as possible and assumes that values near the boundaries are more accurate than values near the middle. All values between start and midpoint or between mid- and endpoint are interspersed values.
    • As long as there are enough periods, interspersed values from the flexible curve can be copied. Interpolation starts from the midpoint.
    • The interpolation of a flexible curve can be applied for 1 <= N <= 2*N phase-in/out periods. If the maximum number of periods is higher no phase-in/out curve will be created.
    • For N=1 the first value of the flexible curve is used, for N=2 the first and the last value is used, for N=3 the first, the last and the midpoint is used. If the interpolated curve has an even number of periods and the underlying flexible curve has an odd number of periods, the midpoint value will be used twice. Each curve has a start, a midpoint and an end value and interspersed values.Some of these interspersed values are copied from the original curve and some are actually interpolated.
    • Number of values to be interpolated per side IC_IPVAL = FC_BVAL - IC_BVAL.
    • FC_BVAL = Number of interspersed (body) values of the original curve per side
    • IC_BVAL = Number of interspersed (body) values of the interpolated curve per side
    • If IC_BVAL > IC_IPVAL then IC_IPVAL periods are interpolated in the middle and the remaining periods are filled with interspersed values from the flexible curve.
    • If IC_BVAL <= IC_IPVAL then interpolation results in interpolated values for IC_IMVAL periods only.

Generation of Phase-Out curves from Phase-In curves: For selected dimension values a phase-out curve is created for all reference products of the new product. Phase-out start date / end date of references = phase-in start date / end date of new product. Identical functions (Linear etc.) are used for phase-in and phase-out curves. Start/end values of the phase-out curve are defined to get phase-in + phase-out = 100%. For a flexible curve this rule is used for every data point where the flexible phase-in curve has values <= 100.

Compare function based phase-in/out curves: Up to 4 curves can be shown after simulation thus allowing the comparison of curves for different parameters. For the comparison a chart and a table view is available. By clicking on a curve in the legend or selecting it in the table view you can save the respective parameters.

Assign user-defined curves: User defined curves can be multiply assigned to values of the launch dimension in the forecast dates table. By clicking on the hyperlink of an assigned curve you directly navigate to the Curves area and the simulated curve is presented.

Assign user-defined curves: User defined curves can be multiply assigned to values of the launch dimension directly using the "save as" function of a new curve.

  • Product assignmts, forecast dates, forecast dates-manual forecast and curves can be uploaded from a .csv file.
  • Launch dimensions are uploaded with the forecast dates upload. A product has only one launch dimension.
  • The separators comma, semicolon and colon are supported in the upload file. Default = semicolon.
  • Via a single upload up to 10000 lines can be uploaded at one time.
  • The upload of forecast dates creates by default a wildcard entry with forecast start date of today for all products contained in the file. By unchecking the flag in the popup this function can be deactivated.
  • With the file upload you can add/modify or delete entries. The file format is identical for all operation types. For deletion it is sufficient to provide the key fields only. Deletion can be disabled via a settings profile.
  • Detailed error messages can be found in the application log in the demand area. There is a direct navigation from the upload popup to the log of the latest upload.
  • The file upload considers permission filters.
  • The file upload is superior compared to the upload with an OData service and is the recommended approach.
  • For the upload of product assignments the following columns are obligatory in the upload file:PRODUCT;DIMENSION2;DIMENSION3;REFERENCE_PRODUCT;WEIGHT;PERIOD_OFFSET;IS_ACTIVE;STARTDATE;ENDDATE. The column names represent the first line of the upload file.
  • The following example for an upload file content loads up two active reference products with weight 50% for product HT_010. The assignment level is Product ID. Caution: You cannot use offsets and validity dates (startdate and enddate) at the same time. When using startdate and enddate make sure that there are no time gaps.
  • No apostrophes have to be used for the upload data.
  • No data for the weighting factor will result in 100%.
  • Anything else than X for IS_ACTIVE will result in an inactive assignment.
  • If dimension2 and dimension3 values are provided, it is checked that planning objects exist for the product and the reference product with the provided dimension values. If this is not the case the assignment is deactivated and a message is written in the application log.
  • You can create planning objects for a new product by copying the planning objects of a reference. Planning objects are created for all planning levels of the selected key figures. By default, all key figures defined for the planning object check in the Settings for Product Lifecycle app are set but can be changed by the planner.
  • The creation of planning objects is only available in normalized planning areas.
  • The upload data for a new product can only contain one reference unequal to the new product. If you have several references assigned you can use a special upload file with only the reference assigned that should be used for copying the planning objects.
  • For the upload of forecast dates the following columns are part of the upload file:PRODUCT;LAUNCH_DIMENSION;DIMENSION_VALUE;FCST_START_DATE;PHASEIN_START_DATE;PHASEIN_END_DATE;PHASEIN_CURVE_NAME;PHASEOUT_START_DATE;PHASEOUT_END_DATE;PHASEOUT_CURVE_NAME. The column names represent the first line of the upload file.
    • The following example for an upload file content loads up two lines with forecast dates for product HT_010. The launch dimension is Location ID (LOCID).

PRODUCT;LAUNCH_DIMENSION;DIMENSION_VALUE;FCST_START_DATE;PHASEIN_START_DATE;PHASEIN_END_DATE;PHASEIN_CURVE_NAME;PHASEOUT_START_DATE;PHASEOUT_END_DATE;PHASEOUT_CURVE_NAME

HT_010;LOCID;HD_DC_AR;20170708;20180101;20180101;Excalibur;20190101;20190331;Phase-Out Linear 22

HT_010;LOCID;HD_DC_JP;20180101;20180101;20180101;Linear;20190101;20190331;Linear - Phase-Out

  • The format for dates is YYYYMMDD (e.g. 20180712) or YYYY-MM-DD (e.g. 2018-07-12). It is recommended to use a tool for editing the file that shows the date format in the way it is stored.
  • Phase-in/out curves are given with their unique name. No technical ID is necessary.
  • No apostrophes have to be used for the upload data.
  • If you upload one line for a product with this template manual forecasting is stopped for the product and all definitions for manual forecast are deleted.
  • With the check box "Generate forecast dates for any attribute value (*)" you can make the upload generate an assignment for any product in the file with forecast start date of today. In this case a forecast is created for all products in the file for which no explicit definitions exist.

  • You can create planning objects for a new product by copying the planning objects of a reference. Planning objects are created for all planning levels of the selected key figures. By default, all key figures defined for the planning object check in the Settings for Product Lifecycle app are set but can be changed by the planner.
  • The creation of planning objects is only available in normalized planning areas.
  • The upload data for a new product can only contain one reference unequal to the new product. If you have several references assigned you can use a special upload file with only the reference assigned that should be used for copying the planning objects.
  • For the upload of forecast dates the following columns are part of the upload file: PRODUCT;LAUNCH_DIMENSION;DIMENSION_VALUE;FCST_START_DATE;PHASEIN_START_DATE;PHASEIN_END_DATE;PHASEIN_CURVE_NAME;PHASEOUT_START_DATE;PHASEOUT_END_DATE;PHASEOUT_CURVE_NAME. The column names represent the first line of the upload file.
  • The following example for an upload file content loads up two lines with forecast dates for product HT_010. The launch dimension is Location ID (LOCID).
  • The format for dates is YYYYMMDD (e.g. 20180712) or YYYY-MM-DD (e.g. 2018-07-12). It is recommended to use a tool for editing the file that shows the date format in the way it is stored.
  • Phase-in/out curves are given with their unique name. No technical ID is necessary.
  • No apostrophes have to be used for the upload data.
  • If you upload one line for a product with this template manual forecasting is stopped for the product and all definitions for manual forecast are deleted.
  • With the check box "Generate forecast dates for any attribute value (*)" you can make the upload generate an assignment for any product in the file with forecast start date of today. In this case a forecast is created for all products in the file for which no explicit definitions exist.

    • For the upload of manual forecasting parameters these columns are part of the file: PRODUCT;LAUNCH_DIMENSION;DIMENSION_VALUE;FCST_START_DATE;SHELF_FILL_START_DATE;PHASEIN_START_DATE;PHASEIN_END_DATE;SMF_END_DATE;BASEVALUE;TRENDVALUE;PHI;SEASONALITY_CURVE;UOM;PERIODICITY. The column names are the first line of the file. SHELF_FILL_START_DATE and SMF_END_DATE can be omitted.
    • The following example for an upload file content loads up two lines with forecast dates for product HT_010. The launch dimension is Customer ID (CUSTID).
    • The format for dates is YYYYMMDD (e.g. 20180712). It is recommended to use a tool for editing the .csv file that shows the date format in the way it is really stored (e.g. notepad).
    • The unit of measure UOM has to be provided and is identical for all lines of a product.
    • The periodicity is optional but identical for all lines of a product (D=days, W=weeks, M=Months,Q=quarters,Y=years).
    • Base value has to be specified, trend value and trend dampening factor phi (0<phi<1) are optional. For the wildcard entry base value can be zero. In this case forecast is performed with the own history of the product and the algorithm chosen in the forecast model.
    • All products in the upload file are marked for manual forecasting. Existing forecast dates for the product, which weren't created for manual forecasting, are deleted.

  • For the upload of curves the following columns are obligatory in the upload file:NAME;CURVETYPE; STARTVALUE; ENDVALUE; FUNCTION; NUMBEROFPERIODS; PERIODTYPE. If a flexible curve with single values is part of the upload additional columns like VALUE 1; VALUE 2 to VALUE N are needed if n data points shall be transferred.
  • The following example for an upload file content loads up three lines with curves.
  • CURVETYPE = 1 = phase-in curve, CURVETYPE = 2 = phase-out curve, CURVETYPE = 3 = seasonality curve.
  • For a flexible and a seasonality curve no start and end value have to be provided. The function of these curves is 8.
  • For seasonality curves a period type (D=days, W=weeks, M=Months, Q=quarters, Y=years) and a start date YYYYMMDD can be specified.
  • These values have to be used for the function: 1 = linear; 2 = square root (superlinear); 3 = cube root; 4 = fourth root; 5 = quadratic (sublinear); 6 = cubic; 7 = fourth power; 8 = single values(flexible curve).
  • The values represent percentages.
  • For product assignments, forecast dates (including Manual Forecasting) and curves a template csv file can be downloaded. It contains the correct headers for the upload.
  • The download of a file is by default done with data. Download without data is possible.
  • For product assignments and forecast dates data is downloaded respecting current filters.
  • All curves of the planning area are downloaded.
  • It is recommended to use the template download and use the downloaded files for the upload.
  • Back up copies can be easily created via the file download.
  • The downloaded files have the unicode format UTF-8.
  • When opening the file with Excel you should use the import function ''From Text/CSV" of the Data tab to bypass the automatic formatting in order to avoid saving wrong formats that lead to errors during upload. We recommend to open the files with other tools like Notepad.

It is possible to add multiple product assignments or forecast dates (option 1) or to edit many existing assignments (option 2) according to filters. This is an additional and optional way to maintain assignments.

The following use cases are supported:

  • Add many new assignments manually via a table maintenance with Excel-like copy&paste functions.
  • Immediate validation of all or single lines of the table.
  • Upload existing csv files with immediate validation and quick correction possibilities.
  • Correct and download existing csv files.
  • Especially the csv file upload is further supported therefore the table maintenance is done in the format of the csv files - existing files can directly be used.
  • Value helps in place for attribute values and dates (right click on dates field).
  • Sorting (right click on column header) capabilities available.
  • For quick troubleshooting in longer lists, there's a filter (Display Rows) available that allows you to display only entries with errors or warnings.
  • Every change can be reverted with the Undo button until the last saved data is reached.
  • With Ctrl + Cursor Down/Up you can navigate to the end or start line of the table
  • Manual Forecast parameters can be simulated in edit mode of Forecast Dates with button Simulate and in detail page of forecast dates with button Add or Edit.
  • Simulation can be done if for the product a periodicity (for the forecast) is defined.
  • By default the forecast horizon is one year starting from the current time bucket.
  • The simulation chart displays the Manual Forecast and the seasonality curve. A table display is available where values of the seasonality curve can be changed.
  • Manual Forecast starts in the time period containing phase-in start date.
  • Manual Forecast can be saved in a key figure configured in the Settings for Prod. Lifecycle app.
  • Seasonality curves are displayed starting from the current period.
  • Instead of a base value a distribution value can be maintained in a different periodicity. From this the resulting base value is calculated. Distribution values are not saved.

You can start the Manage Product Lifecycle and the Manual Forecasting app from Demand Planner area of the Fiori Launchpad. Within the tile, you see the number of product assignments or forecast dates in your default planning area.

In the product assignment overview screen of the Manage Product Lifecycle app you navigate with Planning Object Actions - Create Planning Objects for Selected Assignment to a screen where you define up to 10 additional selection or replace criteria in addition to the selected reference product. Key figures are defaulted from the settings profile of the user. The attributes available for selection have to be in all planning levels of the chosen key figures. If selection criteria are added the key figure selection is frozen. A simulation is available which planning objects should be created for the current selection.

  • Simulation shows for a selected key figure which planning objects are sent for creation.
  • If the function is called from a product assignment you get two table views. One view shows the planning objects that can be copied from the reference. The second view shows the planning objects that were manually added. If the function is called from forecast dates only the second view is present.
  • The simulation does not check if the corresponding planning objects already exist.
  • The simulation and creation does not consider permission filters.
  • If planning objects cannot be copied for a key figure you get a message with the reason.
  • Table entries can be filtered and deleted.
  • Table entries can be downloaded to a csv file.
  • You can create planning objects for all chosen key figures or only for the selected key figure.

In the overview screen of the Manage Product Lifecycle app you navigate with Other Actions - Show Change Log to a list where all changes of the last 120 days to product assignments, forecast dates, launch dimensions and curves are presented. You can also see which action changed data like a manual change in Manage Product Lifecycle, a change of the assignment level, a file upload or the external OData service.

You can filter by the data type, type of change, the action, user and the product.

You can search entries via product, dimension value or the text in the details column.

The log contains information which fields where changed and in case only one assignment was changed the from and to values are available. In addition you see if the changed entry existed or was new.

The change log can be downloaded into a csv file.

Very often the reference products of a reference product are natural candidates for delivering reliable data for new products. On the Product Assignments tab, you can therefore navigate to a hierarchical display of assigned active reference products for a particular reference product including the used weighting factors and select one or several products in the hierarchy as additional reference products if the hierarchy is called in edit mode. The reference product hierarchy view can be opened by clicking on the hierarchy icon within the assignment table.

The hierarchy view posseses a full screen mode as well as zoom, search and node collapse/expand functions.

  • The parameter check aims to determine if the parameter values maintained for Manual Forecasting lead to a forecast that fits to the amount of already sold items of the new product. From historical data of a given key figure a base value, a trend value and a seasonal length are calculated and compared with the maintained values.
  • A deviation is calculated between maintained and calculated base and trend values according to D = (Maintained Value - Calculated Value) * 100% / Calculated Value. If the deviation is greater than the allowed deviation maintained in the settings profile (default allowed deviation = 30%) a warning is raised. The planner can accept all calculated values at once or can accept only some calculated values manually.
  • The check searches subsequent historical data points with positive values starting from the first positive value closest to today e.g. for current period;0;0;0;2;4;6;8;0;7 the time series 2;4;6;8 will be analyzed.
  • Minimum one data point is needed for the calculation of the base value, three data points are needed for the calculation of the trend value. The minimum numbers are configurable via a settings profile.
  • Seasonality check can only be done for a season length of half of the available data points.
  • If enough data points are available to run the forecast with own history (two times the number of the detected season length) a message is raised.
  • For any deviation of maintained and calculated seasonal length a warning is raised. Only in case the maintained seasonal length is greater zero (a seasonality curve is assigned) and the calculated seasonal length is zero an action can be taken. In this case the seasonality curve will be unassigned.
  • In case white noise is detected only base value is calculated.
  • Results of the parameter check are written into the application log.
  • The parameter check can only be executed if a periodicity was chosen for the product.

In the Settings for Product Lifecycle Fiori app settings profiles can be defined and assigned to demand planners for the usage in the Manage Product Lifecycle app. A settings profile tailors the definition process of product assignments and forecast dates according to the needs of a single demand planner.

In addition the assignment level for product assignments is defined per planning area.

If the level1 (Product ID) attribute was changed to a non-root product attribute P2 in the planning area settings, references are assigned on P2 level and forecast dates like phase-in start or phase-in end are defined for values of the P2 attribute. Let us say the level 1 attribute is now product group. In this case product groups are assigned to product groups and for product groups the phase-in and phase-out definitions are done.

The file upload for product assignments and forecast dates works identical to the case where Product ID is the level 1 attribute. There is no change to the column names which means that e.g. values of the product group are written into the column named „PRODUCT".

1 The description of a settings profile can be changed after it was created in the General tab.

2 The usage of notes can be defined in the General tab.

3 The header area immediately reflects changes in description or notes usage. In addition the header are presents administrative data and the assignment level of the planning area for reference products. In read only mode the General tab is hidden as all information is available in the header.

1 You can define if you want to use product assignments. If you say No the product assignment overview and the product assignment tab in the detail screen are invisible in the Manage Product Lifecycle app (MPL).

2 Delete Product Assignments: If set to Yes the planner gets the Delete button and Delete file upload operation type.

3 If you always use a self reference in any creation process an active self reference (with level 2/3 value = *) is generated. You cannot delete a self reference until all other references for the product (+ level 2/3 values) are deleted.

4 If assignments are always active all assignments are saved in status active. Deactivate buttons are invisible in MPL.

5 If Use Offsets is set to No the offset columns in MPL are invisible and all assignments are saved with offset 0.

6 If Use Validity Dates is set to No the columns for validity dates in MPL are invisible and all assignments are saved with empty validity dates (even if the file upload would deliver such dates).

1 You can define up to 3 attributes of the planning area that will be used as additional search and filter attributes. All selected attributes appear in the advanced product search. The product attributes of the selected attributes appear as filters in the Manage Product Lifecycle app. The values of the product attributes are in addition shown as columns in the Manage Product Lifecycle app in case the level 1 attribute is the key field of the product master data type.

2 In addition the list of key figures valid for the simulation of history of reference products in the Manage Product Lifecycle app can be defined. A default key figure can be defined.

1 You can define the default key figures for which the planning objects check and creation in Manage Product Lifecycle is done.

2 You can define if the planning objects check should look for key figure values. Default is Yes.

3 SAP Fiori notifications can be created if no planning objects are found and sent to many planners during a planning objects check. By clicking on the notification you navigate in the Manage Product Lifecycle app where you can download the results in a csv file. Results are available for 30 days.

1 Use-Phase-In: If set to Yes the Forecast Dates table contains by default the columns phase-in start date / phase-in end date and phase-in curve. These columns are hidden by default if phase-in should not be used.

2 If phase-in start always equal to phase-in end only phase-in start date is visible in MPL, no end date needs to be maintained. No phase-in length can be maintained and the corresponding section is hidden.

3 If phase-in settings are allowed for any attribute value you can maintain phase-in dates for wildcard entries.

4 Use-Phase-Out: If set to Yes the Forecast Dates table contains by default the columns phase-out start and end date and phase-out curve. These columns are hidden if phase-out should not be used. In case phase-in and phase-out are set to No forecast dates overview and forecast dates tab in detail screen are invisible.

5 If phase-out start always equal to phase-out end only phase-out start date is visible in MPL, no end date needs to be maintained. No phase-out length can be maintained and the corresponding section is hidden.

6 If phase-out settings are allowed for any attribute value you can maintain phase-out dates for wildcard entries.

To be able to set a phase-in or phase-out length the flag Use Phase-In or Use Phase-Out is set to Yes.

1 First you have to select an attribute of the product master data type of the planning area.

2 You can define phase-in/out lengths for attribute values of the selected attribute.

As periodicity you can choose days, weeks or months if they exist in the planning area.

With the phase-in/out length missing start or end dates are calculated. This means if phase-in/out lengths are defined it is sufficient for the planner to specify only a start or a end date. The system adds/subtracts the respective days to/from the start or end dates. For weeks the number of days is length * 7, for months the number of days is length * 30.

1 Use Curves: If set to No the curves tab and all fields related to curves are hidden.

2 Create and Delete User-Defined Curves: If set to Yes the planner can save and upload phase-in/out curves.

3 Show and Use all User-Defined Curves: If set to Yes the demand planner can view and assign all existing phase-in/phase-out curves. Otherwise the planner is restricted to own curves.

4 Delete Forecast Dates: If set to Yes the planner gets the delete button and the file upload operation type.

5 If definitions for Any Attribute Values are mandatory in any creation process a wildcard entry is generated with a forecast start date of today. You cannot delete a definition for any attribute value (wildcard entry) until all other forecast dates for the product are deleted.

The administrator can define which attributes of the planning area can be used as launch dimension and which attribute is the default launch dimension. If the demand planner adds a new product via the Manage Product Lifecycle app only the launch dimension defined here can be selected in the value help and the default launch dimension is preset. During mass upload launch dimensions in settings profiles are not considered.

For each selected launch dimension, except for the attribute chosen as level 1 attribute of the planning area, a preferred value can be specified. If preferred values are defined the planner gets a filter for preferred values in the Forecast Dates Tab in Manage Product Lifecycle app. In addition these values are put on top in the value helps and are set by default when creating new entries for the respective launch dimension.

Default values for the unit of measure and the planning periodicity of the products and for the periodicity used for the distribution value can be set. Without these default values the base unit of measure of the product and the periodicity mostly used in forecast models is set in Manage Product Lifecycle.

It can be defined if shelf fill values including shelf fill dates can be set and saved in the simulation.

For the simulation a default forecast horizon can be defined. If nothing is defined simulation runs for one year.

Key figures available for the save of the simulated forecast can be selected.

For the interactive parameter check of Manual Forecasting some settings can be defined.

If the deviation between the maintained and calculated base and trend values is bigger than the allowed deviation the check will raise a warning. The default allowed deviation is 30%. Any deviation in seasonality lengths will trigger a warning.

For base value and trend value a minimum number of data points is needed to achieve reliable results. The default number for base value is 1 and the default number for trend value is 3. If the number of available data points is lower than the defined minimum number no check is executed.

A default key figure for the check can be defined.

1 Up to 100 users can be assigned to a settings profile. A user can only be assigned to one settings profile in one planning area. If more than 100 users should be assigned please work with user groups.

2 Up to 100 user groups can be assigned to a settings profile. A user group can only be assigned to one settings profile in one planning area.

The direct assignment of a user overrules the assignment of a user group where the user belongs to.

If a user is assigned to multiple user groups and the multiple user groups have different assignments of settings profiles, the settings profile is chosen that was changed latest.

Visibility and Editability of References and Forecast Dates

  • With help of an external master data type you can view references and forecast dates in a master data sheet of the IBP Excel Add-In. Data changes are not possible this way.
  • For the definition of the external master data type you can take advantage of the external data sources SMD_FCSTASSIGN and SMD_FCSTASSIGN_V2 for references and SMD_FCSTDATES and SMD_FCSTDATES_V2 for forecast dates.
  • For the definition of the external master data type you need to create appropriate attributes.
  • All key fields of the external data types have to be added as attributes to the external master data type. For assignments this is planning area, product, dimension2, dimension3 and reference product. For forecast dates this is planning area, product and dimension value. This means that always data for all planning areas and dimensions is shown.
  • For planning area you need to use an character attribute of length 10.
  • For product, reference product and the dimensions you use char. attributes of length 60.
  • Validity dates of product assignments can be shown using the fields startdate and enddate.
  • Dates are defined as Character 8 values and are shown in the format YYYYMMDD.
  • Phase-In curve names and phase-out curve names cannot be shown with SMD_FCSTASSIGN and SMD_FCSTDATES.
  • The external master data type with its attributes has to be assigned to your planning area.
  • If you assign attributes of the external master data type to a planning level you will be able to filter the output according to these attributes in the master data workbook.
  • You need to create new attributes for external master data types. One attribute can only be used in one master data type in a planning area.
  • In the Excel add-in, you may choose Web Apps - Manage Product Lifecycle from the About group of the SAP IBP ribbon, and select Product Assignments or Forecast Dates and you go to the respective edit mode. The planning area set in the add-in is taken over. Further filters are available in the Manage Product Lifecycle app. With this external master types are not needed any more to visualize product assignments and forecast dates. There is currently no navigation to manual forecasting parameters.

External Data Sources SMD_FCSTASSIGN_V2, SMD_FCSTDATES_V2, SMD_FCSTASSIGN_DATES

  • With the external data source SMD_FCSTASSIGN_V2 validity dates are shown in a formatted way YYYY-MM-DD. Additionally the product key fields have length 40 to make it compatible with the delivered product attribute PRDID. The following fields are part of the data source: PAREA(key, char 10), PRODUCT (key, char 40), DIMENSION2 (key, char 60), DIMENSION3 (key, char 60), REFERENCE_PRODUCT (key, char 40), WEIGHT(Integer), STARTDATE(char 10), ENDDATE(char 10), PERIOD_OFFSET(Integer), IS_ACTIVE(char 1), LAUNCH_DIMENSION(char 80), PRODUCT_DESC(char80), REFERENCE_PRODUCT_DESC(char 80), DIMENSION2_DESC(char 60), DIMENSION3_DESC(char60). Even if level 1 attribute is set to e.g. Product Group the header descriptions remain PRODUCT and REFERENCE_PRODUCT.
  • With the external data source SMD_FCSTDATES_V2 dates are shown in a formatted way YYYY-MM-DD. Phase-In and Phase-Out curve names can be shown. Additionally the product key fields have length 40 to make it compatible with the delivered product attribute PRDID. The following fields are part of the data source: PAREA(key, char 10), PRODUCT (key, char 40), DIMENSION_VALUE (key, char 60), FCST_START_DATE(char10), PHASEIN_START_DATE(char10), PHASEIN_END_DATE(char10), PHASEINCURVENAME(char 40), PHASEOUT_START_DATE(char10), PHASEOUT_END_DATE(char10), PHASEOUTCURVENAME(char 40), DIMEN_VALUE_DESCR(char60). Even if level 1 attribute is set to e.g. Product Group the header description remains PRODUCT.
  • With the external data source SMD_FCSTASSIGN_DATES you can display product assignments and forecast dates in one table having the above key and non-key fields combined.

General Limitations

  • Only products (in general the attribute values of the attribute defined as level 1 attribute) are replaced in the selection of a forecast runIf e.g, forecast is run for product = ProductNew, location = Location1, customer = customer1, the forecast engine reads data forproduct = ProductOld, location = Location1, customer = customer1 in case ProductOld is defined as a reference product for ProductNew
  • Product references are only taken into account by forecasting if the forecast level contains the level 1 attribute e.g. if forecast is done on product group level, no product references can be taken into account if product ID is the level 1 attribute.
  • Manual forecasting parameters are not considered on aggregated level even if the flag "Aggregated planning for Phase-In/Phase-Out" is set in the forecast model.
  • Product lifecycle definitions can only be made for the base version.
  • The attribute values of products (level 1), level 2 or level 3 attributes or launch dimensions cannot be longer than 60 characters.
  • Attribute permissions are not considered.
  • Write permission filters are not considered.

Additional Limitations in Demand Sensing

  • Demand sensing does only support level2 and level3 attributes being part of the location or customer master data type. Level 2 and level 3 attribute can be from the same master data type.
  • Simulation of demand sensing containing reference product assignments in Excel does not consider manual changes as demand sensing always reads data of reference products from the data base.
  • Demand sensing can only consider lifecycle settings if the level 1 attribute is the attribute with business meaning Product Identifier.
  • Setting "Use Reference Product History after Phase-in End Date" in the planning area settings of the Settings for Product Lifecycle app is not considered.

Trouble-Shooting

  • Sometimes forecast is not generated or generated differently to expectations. Here is a short list of most common misunderstandings that can cause irritation.
    • Check that level 1 (mostly product ID), level 2 and level 3 attribute are part of the forecast level if product assignments should be considered.
    • Check that the launch dimension attribute (or root of launch dimension attribute) is part of the forecast level if forecast dates should be considered. If the master data type containing the launch dimension has more than one root attribute the launch dimension itself has to be part of the forecast level.
    • Forecast engine needs planning objects for the new product for the forecast key figure and for the source key figure for forecasting. If this is not the case no forecast is generated.
    • Forecast engine is only applying curves to the forecast key figure in the forecast horizon.
    • If forecast dates are defined for some values of the launch dimension no forecast is created for all other values of the launch dimension. Maintaining wildcard entries avoids this.
    • Deleting all forecast dates of a product does not delete the launch dimension of a product. Delete function in the detail page of Manage Product Lifecycle deletes all data related to a product inclusive the launch dimension.
    • If you want to stop manual forecasting for a product you have to delete the launch dimension for the product in Manual Forecasting app. This may be needed e.g. if you want to define phase-out settings for a product.
  • Sometimes descriptions of attributes are not shown in the Manage Product Lifecycle app.
  • Check that the description attribute of the product, the launch dimension or the level 2/3 attribute is part of the planning area.
  • Set the corresponding business meanings like Product Description or Location Description or others in the planning area configuration.
  • If no appropriate business meaning is available specify in the master data type which attribute is the description attribute for respective attribute.
  • If you go to Settings for Product Lifecycle app and there to Planning Area Settings and set "Allow Master Data Values in Assignments" to Yes you can show the description of the attribute values if only the description attribute is specified in the master data type.
  • If you missed to do the appropriate settings and correct your settings later you can get the correct description using the Synchronize function in Manage Product Lifecycle.
  • Sometimes product assignments cannot be saved in active status
  • Check that planning objects exist for product and reference product (and the level2/3 attribute values).

Save progress to your learning plan by logging in or creating an account

Login or Register