Understanding Supply Chain Attributes for Better Forecasting
What are attributes of a supply chain?
Attributes are properties specific to an object. In a supply chain, like anything else, everything that you deal with has attributes, such as:
- Customers: priorities, size, repeat business potential
- Products, raw materials and WIP: speed, texture, grade, quality, taste, shade, color, size
- Equipment: precision, maintenance, cost, speed, output quality, carbon emission
- Suppliers: qualification, quality, service level, cost region
- Processes: customer approved, output quality, yield
- Regions: labor laws, stability, price, regulations, import/export restrictions, transportation, earthquake prone
There are more: objects such as tools, cost restrictions, packaging skill levels, storage temperatures, etc.
Such attributes are required in order to plan and peg products to the right order of a supply chain. Unlike what is commonly believed, attributes are not just for the finished good! Attributes appear at every stage of sourcing, making and delivery. For example, one needs to use a specific qualified supplier, process and product specifications for an end user. In the absence of attributes, most supply chain vendors use multiple instances of the same BOM and routings to represent customer-specific options. The problem with this approach is that even if you have 20 different attributes (in reality, there are many more when you consider the entire supply chain), the number of combinations of SKUs and bills of materials will be around 2.5 quadrillion! Now you know why some systems run so slowly even if they use in-memory computation.
Why do we need Attribute Based Planning in supply chains?
With Attribute Based Planning (ABP) capability, a generic product (BOM) is defined and then all the attributes are dynamically attached to the generic BOM, based on the specifics of the demand. This reduces the complexity of the model and memory requirement by orders of magnitude and keeps the SKUs to the minimum. It also reduces the size of master data significantly and its maintenance.
One other very important use of Attribute Based Planning is that, as the structure of your supply chain is constantly changing (e.g. acquisition of new companies, new products, new customers or new suppliers), users can simply add or delete attributes as needed. The system would then automatically take into account the new attributes into its search strategy every time it runs. In other words, attributes take the role of dynamic constraints that can be added or removed by the users as the business changes. Thus, the system is constantly adapting itself to its environment. The latter is done by using machine learning algorithms that can understand the changing patterns in the supply chain and update the attributes accordingly. For example, efficiency of an equipment going down over time, then its attributes are adjusted accordingly. This is what we refer to as self-correcting models of the supply chain in order to maintain a true digital twin of the supply chain.
In summary, attributes make the system flexible, adaptable, minimize CPU and memory requirements, make maintenance of master data much easier and, most importantly, enable representation of the supply chain accurately. They represent the behavior of the supply chain and not just the structure thereof. The use of Attribute Based Planning in digitalizing supply chains, make it a lot easier and faster for companies who are on their journey to form digital twins of their supply chain and perform autonomous planning.