Supply Chain Excellence
Mathieu Pourqué
Forecast Accuracy Champions
Aroma Performance: Marie Pereira
Spp: Cecile Bailly
Novecare: Jean-Baptiste Cercueil
Peroxides: Benoit Hornyak
Silica: Philippe Deloro/ Jose Pablo Leyva
Special Chem: Anne Vanesse
Soda Ash & Derivatives: Fabien Piret
Technology Solutions: Aleksis Parfens/ Laure Imberti
O&G: Charles Aymes
General Description |
|---|
The KPI Forecast Accuracy (FA) evaluates the ability to get a visibility on customers' demand in terms of quantity. As an input of S&OP process, reliable forecasts constitute the major way to improve the customer satisfaction via an optimized planning of operations. The Forecast Accuracy in this dashboard is calculated based on the Final Forecast and the Gross History. Check definitions by GBU here For a particular month, the Forecast Accuracy is calculated for the last 5 months forecast (M-1 ... M-5); allowing the possibility to evaluate the accuracy of the forecast with reference to several different forecast periods. Forecast Accuracy M onth-2 is the time between produced forecast and the Gross History (FA June M-2: means forecasts done in April, as shown in the illustration below) The refresh of the KPI is made every month on the 7th, including the full history. It is available for all GBUs currently using Dynasys or APO as Demand Planning systems: Aroma Performance, Composite Materials, Novecare, Peroxides, Silica, Soda Ash&Derivatives, Special Chem and Technology Solutions. |
Source Data in the GSCD
GBU | Source | Gross History | Final Forecast |
|---|---|---|---|
Aroma | Dynasys | Demand History ETA | Last validated Forecast ETA |
Silica | Dynasys | Demand History ETA | Last validated Forecast ETA |
TS | Dynasys | Demand History ETA (Sales Orders volumes in ETA) | Sales Team Forecast ETA |
Novecare | Dynasys | Demand History ETA (Last Requested Delivery Date) | Last validated Forecast ETA |
SA&D | Dynasys | Actual Shipped Qty in ETD | Last Validated Forecast ETD (Unconstrained Demand validated after demand review meeting) |
Peroxides | Dynasys | NA: Shipped History ETD Others: Demand History ETA | NA: Last validated Forecast ETD Others: Last validated Forecast ETA |
Spec Chem | Dynasys | Demand History ETA | Last validated Forecast ETA |
Please note that O&G data can be found in GSCD but is not capturing all regions and markets; Data taken into account in the calculation is ETD.
FA of Spp is available on E2E value chain dashboard, calculated ETD, but not weighted as per the group definition.
| Aggregation levels: what is the purpose of each level? | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Purpose: what can we really measure with each dimension/aggregation level in order to understand who is the Key Responsible to track this KPIs. Roles & Responsibilities: Responsible is considered the person who will look after this KPI in a monthly basis and Accountable is the person whose decisions can leverage a better insight and/or opportunities to improve the KPI Timeframes (Lag): they depend on how the business is structured and their standard leadtimes, eg. if we talk about a business mostly MTO driven with total replenishment lead times that last around 3 months then the operational purpose is at M+4, if instead it is a business mostly MTS driven M+2 will give an insight on operational purposes. Overall we can say that: M-n: Operational purpose, M-n+2: Procurement purpose and M-n+5: Workload/Contract purpose
|
| Glossary |
|---|
Figures Final Forecast = Total on quantity forecast to be shipped. Final Forecast is the forecast validated after the Demand Review (=Unconstrained Forecast)Gross History = Customer Demand = Last Customer RequestCheck the above definitions by GBU here Dimensions Distribution Channel = Sales Distribution ChannelGBU Prod. Family = by default the Product Family, but also the Product Hierarchy, the GBU Material Group, or the Forecasts FamilyGBU Zone = Mini-Zone = Group of Countries, Specific by GBUPackaging Type = used on a materialProduct Group (PGMI) = Group of products from DynasysProduct Line 00 = Attribute of a material. There are 6 differents product line levels, corresponding to differents aggregations of productsSales Rep = Sales Representative from Dynasys Sales Group = DFU ownerShip-to KA = Ship-to Key Account, final accountTactical Material = Concatenation of Product Hierarchy and Packaging |
WMAPE: Improvement of Forecast Accuracy Weighted formula |
|---|
The Forecast Accuracy (MAPE) measures the accuracy of the forecasting figures. In Solvay the figures are based on the Final Forecast validated in the Demand monthly Reviews vs. the Gross History. Forecast accuracy at the SKU level is critical for proper allocation of resources. Weighted Forecast Accuracy (wMAPE) is a variant of MAPE in which errors are weighted. In the past we used to weight the values of actuals, which did not penalize the over forecasting cases. To overcome this situation, now we weight using the Total of Actuals and Forecasted volumes by the Aggregation Level (Material, SREP or product). NEW Weighted Forecast Accuracy Formula WMAPE = MAPE x Weight Forecast accuracy (MAPE) = (1- Abs((Final Forecast - Gross History)/Gross History))) for each line by aggregation level Weight = (Final Forecast + Gross History)/(Sum( Final Forecast + Gross History)) for each SREP, material or product  |
| Forecast Error: Bias & Tracking Signal (NFM) | |
|---|---|
Forecast error measurement should serve a purpose. From a demand forecasting perspective, the purpose is to understand the planning process capability and identify products that are systematically the biggest error contributors and set actions to offset the effects of them. | |
The objective is to introduce a framework for using Solvay’s forecast error metrics according to a structured process:
|
|
Measure Forecast Deviation Forecast Bias % (PE) = (sum(Final Forecast)- sum(Gross History))/Sum(Gross History) x 100 When to use: The conversion of forecast error into percentages, allows for comparison across products with different magnitudes of demand volumes. It is also used to indicate underforecast or overforecast. When the computed error is greater than 100% in either direction, the measurement result should be shown as it appears. | Diagnose the forecast error and root causes: Normalized Forecast Metric = (Final Forecast - Gross History)/ (Gross History + Final Forecast)) When to use: as a diagnostic measure for finding forecast line items with the most significant forecast bias in a diverse group of products. It can be used regardless of their scales of volume or units of measure. |
| |
| Technical Documentation | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Source: DYNASYS / APO (WP1) / APO (PF1) - QV_BW_QRY_MVDYN11_0005Update: Monthly full reload (history included) on the night of the 6th and the 7th (minor corrections in Gross History may still occur from one month to another)Technical BW Documentation: BW - DP - Forecast Accuracy (Core Query)Naming convention by Source System (Dynasys vs APO)
Dimensions: Global Overview of Dimensions
|
Not able to find the solution? Contact SBS Support.