Uncertainty in supply chain elements

Support from the Finance and Accounting group within your organisation is vital for the effectiveness of Supply Chains. A step in gaining support is through segmenting the elements of your organisation’s supply chains into categories for improved management.

The main elements within a supply chain are customers, finished goods inventory and service parts (when applicable), suppliers and purchased items. Each element has its own methodology for segmentation, enabling a structure to be designed for better understanding.

Inventory management steps

Inventory Management comprises Inventory Policy, Inventory Planning and Inventory Control. The previous blogpost discussed some aspects of Inventory Policy.

This blogpost considers segmentation for Inventory Planning, through identifying the ‘patterns of inventory movement’. This approach differs from that undertaken by Marketing in defining its market segments.

Safety stock is the inventory volume carried in excess of the cycle inventory required to ensure that customer demands can be met. Safety stock is calculated using the forecast ‘error’ (a statistical term to identify the difference between planned and actual situations).

The range of forecast errors for an SKU enables a logistician to identify different demand patterns. Therefore, multiple SKUs with similar demand patterns can be collated together into a category for common decisions to be made about planning and the level of control required.

Unpredictable and variable demand for certain SKUs is imposed by Marketing actions that are seen to be critical to grow a business. These actions include product line extensions (that provide the ‘long tail’ of inventory), product promotions and new product introductions.

An approach to inventory management is therefore required that identifies those SKUs with a more stable demand and which require less planning and analysis effort and buffers of inventory and idle capacity. This enables sufficient resources to be deployed to manage the unpredictable and variable demand items.

Co-efficient of Variation

The limitation of using traditional ABC categories for analysis is that all items within a category are treated the same, as only annual sales are considered and SKUs categorised using the Pareto (80/20) rule.

To overcome this limitation, use the Coefficient of Variation (CoV) as the measure of volatility in demand and to assess the predictability of a demand pattern, that is, how well it can be forecast.

The calculation is: Coefficient of Variation = Standard Deviation / Mean. For calculation, this requires historical data of annual units sold, in 52 week buckets, listed by SKU in descending order. The lower the CoV, the more consistent is the sales pattern. If a CoV >1.0, the variations in demand are high and therefore care should be taken in using common statistical techniques.

The next step is to plot the data for each SKU on a scatter chart. Place the annual sales volume on the Y axis and the CoV on the X axis. This provides a picture to develop a strategy for SKUs held in inventory.

But this approach leaves a logistician short of answers concerning the ‘what and why’. To assist using the output from CoV for inventory decisions, Tom Rafferty of Supply Chain STO P/L developed the coefficient of variation for management (CoVM^©) model.

By combining the calculation for Category and Class by SKU, the result can be placed into a table. Categories are on the Y axis and Class along the X axis.

Using this structure, it is more likely that, for example, cells Aa, Ba and Ca, have similar patterns (and so will cells Ab, Bb, Cb) and therefore each Class group can be managed in a similar manner.

The inventory groups are identified as STEADY, VARIABLE, ERRATIC, IRREGULAR, LUMPY and DEAD. The example below illustrates seven SKUs that provide Category A. However, the CoV calculation shows that when considering Category/Class, two are STEADY, three are VARIABLE and two are ERRATIC.

Actions by Category/Class

The actions required by a logistician for each Category/Class are based on the similar pattern of demand.

All SKUs that appear in Class ‘a’ have a STEADY demand, with little variation and actual sales close to forecast. Inventory is managed using a Tracking Signal. To calculate, divide the cumulative variation for the periods under review by the standard deviation. The acceptable tracking signal will be less than 4.0.

SKUs in the VARIABLE Class will have a varied forecast ‘error’. Use a Tracking Signal of more than 7.0 as the trigger for a review. Use applicable safety factors (which provide different safety stock requirements), especially in Category/Class Ac and Bc.

The ERRATIC class may have SKUs with high annual sales, but which vary by month or season. Sales are often to a small group of customers, so Logistics must work with Sales to understand the drivers of demand.

The SKUs in Classes IRREGULAR, LUMPY AND DEAD provide the Slow and Obsolete (SLOB) group. SKUs in each Class have not registered any sales for 6 of the preceding 12 months and consume much of the inventory holding costs. SLOB items cannot be forecast using traditional time series based forecasting techniques. Instead, use a Poisson probability distribution that recognises the small and infrequent demand patterns.

For SKUs in the IRREGULAR Class, consider different ways to satisfy customer demand, such as Outsourcing production to a smaller company. If the item is imported, often with minimum order quantities, review the gross margin and customer service level.

The LUMPY Class can provide up to 70 percent of all SKUs, but represents 5-7 percent of sales (the long tail of inventory). This Class is inhabited by line extensions that Marketing are convinced will be a ‘winner’. Logistics has a challenging task to reduce the number of SKUs in this group and to stop the introduction of more, through identifying their costs to launch.

The DEAD Class has not sold an item in the past 12 months. Except if they are ‘insurance’ service parts, the SKUs should be eliminated from the sales catalogue.

Managing Service Parts inventory is a often a dedicated function. The segmentation of service parts is based on the criticality of each part to the business. For inventory purposes, the items generally fit into the LUMPY Class.

Structuring your organisation’s inventory using CoVM provides a professional approach to reaching agreement with Finance that inventory is not a cost to be reduced but an asset to be planned. If the outcome of the plan costs too much, then the Sales and Operations Planning (S&OP) meeting is the venue for addressing the business model.