Strategic Decision Paths for Supply Chain Planning in Retail

[Manaswi_Dubey1]

Summary

This accelerator provides businesses with a general guidance on making architectural decisions regarding the business capability of Supply Chain Planning specifically focusing on forecasting & replenishment business processes for the Retail Industry. It focuses on SAP Customer Activity Repository applications bundle (CARAB) based solutions- Unified Demand Forecasting & Replenishment Planning in addition to Retail Industry cloud solutions such as SAP Predictive Replenishment. The accelerator aims to provide insights into the optimal solution paths based on different customer landscapes and business needs & diverse customer situations. 

The decision accelerator provides guidance for:

• Existing CARAB customers: Considerations on the transitioning to the use of forecasting & replenishment capabilities of CARAB.
• New customers: Guidance on Retail Industry Cloud capabilities around the replenishment process.

Business Requirements

• Assortment Planning: Forecast information is needed to plan retail assortments in alignment with the financial plan and in season stock projections.
• Promotion Optimization: Leverage a robust forecast data to adjust promotional attributes and periods as per financial KPI’s.
• Merchandise Allocations: Demand plan is the base for merchandise push scenarios.
• Replenishment Planning: Using the demand plan, store replenishments need to be executed to fulfill merchandise pull scenarios.
• Reduce manual repetitive tasks, managing critical demand and inventory situations while ensuring cost optimal replenishments.

Business Capability mapping

The SAP Reference Architecture provides the guardrails for conducting a business-driven exploration of the underlying business capabilities and mapping them to SAP solution offerings. It consists of the SAP Reference Business Architecture (RBA) which is a standardized content for the business architecture of any enterprise & the SAP Reference Solution Architecture (RSA) which is the recommended mapping of the business capabilities to SAP's solution portfolio.

RBA is structured along four Enterprise Domains- Products and Services, Customer, Supply and Corporate. All enterprises can be structured along these four Enterprise Domains. Below the Enterprise Domains, RBA consists of the Business Capability Model (BCM) which organizes enterprise’s business capabilities. The model can be used by enterprises for evaluation & identification of improvement areas. Delving further into the business capability of Supply chain planning will provide us the mapping of the solution capability from SAP’s solution portfolio.

Solution Options

Before we dive deeper into the solution options, it would be worthwhile to understand the environment of the respective solution capabilities.

The below diagram provides a quick overview of SAP CAR, its modules & the consuming applications which together are termed as SAP Customer Activity Repository Applications Bundle (CARAB).

A brief overview of the various modules & consuming applications within SAP CAR-

• POS Data Transfer & Audit (POSDTA)- Collects, audits & transfers POS data.
• Multichannel Transaction Data Management- Define sales channels & collect wide variety of data types.
• Demand Data Foundation- Reusable data layer that provides consuming applications with planning, analysis, and forecasting capabilities.
• Unified Demand Foundation (UDF)- Provide demand modeling and forecasting services for SAP applications driven by demand prediction.
• Promo offer Management- Basic FIORI based UI apps to enable creation & maintenance of offers to be executed by Omnichannel Pricing & Promotion.
• On- Shelf Availability- Provides insights on the current and past on-shelf availability in the stores.
• Distribution Curves- Calculate distribution curves for different consuming applications.
• Location Clusters- Group locations into clusters to be used by other consuming applications.
• Product Attributes- Manage and maintain product attributes for a selected product hierarchy.
• Omnichannel Pricing & Promotions- Service to ensure correct & consistent sales price across all sales channels.
• Inventory Visibility & Omnichannel Article Availability & Sourcing- Provide real- time inventory, article availability & sourcing information across all sales channels.
• Assortment Planning- Plan product to sell, locations & strategy across all channels.
• Merchandise Planning- Plan sales, OTB, margin, inventory, receipt flow, etc. for merchandise.
• Promotion Management- Leverage data & predictive algorithms in CAR to plan & manage promotions for all sales channels.
• Allocation Management- Plan & execute the distribution of products from DC to stores.
• Replenishment Planning- Intraday replenishment with cost optimum ordering & sustainability features.

SAP recommends the below solution options to meet the requirements around forecasting & replenishment of merchandise across retail stores & DC’s-

• Utilize the Unified Demand Forecast (UDF), a module of SAP CAR which provides demand modeling and forecasting services for applications driven by demand prediction. UDF generates a demand plan considering various predefined Demand Influencing Factors (DIF’s), demand models which incorporates certain what- if scenarios & is also capable of considering external forecast information from 3rd party systems.
• The output from UDF can be adjusted as per user analysis & also utilized by other CARAB consuming applications such as SAP Replenishment Planning (RPL) to plan store replenishments.
• SAP Predictive Replenishment (PRP) is an industry cloud solution hosted on SAP BTP (Business Technology Platform). It focuses on replenishment planning for retail & makes ordering of retail products more efficient. This industry cloud solution leverages AI to manage inventory efficiently and accurately for an optimized multi- level supply chain. The solution uses machine learning based forecast and considers business targets, constraints, and cost to auto- determine order quantities with lowest expected costs.

The short- term approach here is that the forecasting capabilities are handled by Unified Demand Forecast whereas the replenishment capabilities are handled by both SAP Replenishment Planning for store replenishments & SAP Predictive Replenishment for DC replenishments. The demand plan from UDF can also be shared with Forecasting & Replenishment (SAP F&R) for both DC & store replenishments. This is more of a short-term solution option and should be considered bearing in mind the restrictions around SAP’s maintenance strategy & PAM information for F&R. The long term vision would be to leverage SAP Industry cloud for Retail solution capabilities for both forecasting & replenishment.

Characterization of the Solution Options

To understand the solution approach, we should take a deeper look into the in-built capabilities of each of the applications which enable a process as critical as forecasting & replenishment of merchandise in the retail industry. Understanding the solution capabilities, their integration touch points & sustainability features will be key to align the solution approach to a customer’s strategic initiative.

Unified Demand Forecast (UDF) is an advanced statistical forecast using machine learning algorithms. It uses multiple forecasting methods to provide demand forecasts to multiple CARAB applications such as SAP Promotion Management for Retail, SAP Assortment Planning, SAP Allocation Management, SAP Replenishment Planning, SCM applications such as SAP Forecasting & Replenishment (SAP F&R) and industry cloud solutions such as SAP Predictive Replenishment. It uses master & transactional data to analyze historical demand and generate a machine learning demand model for each product- location. Demand influencing factors are considered & weighted based on historic observations following supervised learning. The trained model is scored to produce forecasts considering future values of influencing factors. UDF uses Bayesian statistics to fill in gaps of knowledge in case historical data is not available and then calculates the forecast. This forecast information is at a granular level with a flexible multichannel model and can be intraday as well to handle fresh produce.

SAP Replenishment Planning (RPL) is a store replenishment application of CARAB which uses the demand plan from UDF and offers cost optimal store replenishment order proposals. It evaluates in parallel multiple order proposals across multiple parameters such as less waste (for fresh products), loss of sales, fuller shelves, optimal cost, etc. and determines the best order proposal by balancing all these attributes. Such a balance can be different based on the product category and merchandising objectives. These can be maintained to be considered during the best order proposal selection process. The RPL application aims to reduce cost by simulating the impact of replenishment settings over time on KPI’s such as stock outs, revenue, spoilage, etc.

SAP Predictive Replenishment (PRP) is an industry cloud solution focusing (right now) on DC replenishments. As a native cloud solution and part of SAP's Industry Cloud solution portfolio PRP is hosted on SAP BTP and it provides business with the required agility to run their operations both optimally and automatically. The first focus of the solution is on distribution center replenishment. The concept of cost optimal ordering has been leveraged in PRP to further optimize order quantities as per vendor restrictions. An amalgam of both private & industry cloud solutions could be the use of RPL & PRP in conjuncture thereby providing a true multi- echelon demand planning for store & DC replenishments. The store order proposals from RPL are aggregated and used as input into PRP for DC demand planning.

Decision Tree

A decision tree to determine the target application architecture is illustrated below

At the start of a Supply Chain Planning transformation initiative at a retail customer, it would be worthwhile to first evaluate if SAP CAR is one of the applications in the application architecture.

Existing SAP CAR customers might be using it only for its POSDTA capabilities. If so, we should investigate the applications being used for the business areas of- Demand Planning (forecasting) & Supply Planning (replenishment). If the customer has SAP F&R, it would be leveraged for both business areas. In such cases, a short-term recommendation would be to move the forecasting process to SAP UDF & retain the replenishment process in SAP F&R. Mid- term, the customer should plan to move the store replenishment to SAP RPL while the DC replenishment can move to SAP PRP. This will enable the customer to rationalize applications nearing end of maintenance & would also be a good starting point for the business to leverage on the ML based capabilities of SAP UDF & also move to sustainable solutions such as SAP RPL. The eventual goal would be to move to Industry cloud solutions such as SAP PDP & SAP PRP.

If the customer is using non- SAP applications for both (or either) of the business areas, the short term recommendation would be to make use of SAP UDF for forecasting, SAP RPL for store replenishments & SAP PRP for DC replenishments. The long-term recommendation would be to eventually leverage Industry cloud solutions of SAP PDP & SAP PRP for forecasting & replenishment respectively. The decision path for net new customers would also be the same.

In case the customer is using SAP UDF for forecasting & SAP RPL for replenishment capabilities, they are already on the short-term recommendation. The long-term vision should be to move the industry cloud solutions.

For retail customers using non- SAP applications other than SAP CAR, the recommendation would be to onboard them to SAP CAR with the short- term approach & eventually move to the long- term approach of industry cloud.

Conclusion

Customers using SAP F&R for the supply chain planning of merchandise to retail stores & DC’s should evaluate to move the 'F'orecasting piece to SAP UDF on CAR using which they will be able to leverage on the machine learning capabilities while also building on the Bayesian statistical models of SAP UDF. The ‘R’plenishment part can still stay in F&R for the short term but eventually should move to SAP RPL & SAP PRP in the interim. In the long run the intention is to leverage on the industry cloud solutions. With this approach, customer can ensure that the supply chain planning business capability is supported end to end by native cloud-based solutions.