Inventory stratification is the process of classifying items based on predetermined factors related to a company’s business environment and goals. The methodology organizes inventory items and stock keeping units (SKUs) into categories in order to optimize working capital. Being revenue-driven, inventory stratification ranks items on their profitability and how quickly they sell. Fast-moving items, even if they don’t earn much profit, are positioned ahead of items that are highly profitable, yet sell at a slower pace. Those SKUs that are less lucrative and do not sell in a reasonable amount of time are removed from the inventory mix. The key to inventory stratification, therefore, is attaining an optimal balance of carrying just enough slow- and low-selling inventory to meet customer demand without burdening cash flow.
Inventory stratification is grounded in Pareto’s law, which, according to the APICS Dictionary, states that a small percentage of a group of items represents the greatest impact or value. In the philosophy, ABC classification is used to create three distinct groups of SKUs, labeled A, B or C. The APICS Dictionary explains that A items usually comprise 10-20 percent of the total number of items, but as much as 50-70 percent of the dollar volume; B items account for 20 percent of the total and 20 percent of the dollar volume; and C items typically make up 60-70 percent of the total number of items and 10-30 percent of the dollar volume.
This type of categorization is important because, for most businesses, enormous amounts of employee time and energy are consumed by inventory management. A items may deserve these efforts because they are so critical. However, B items often are better handled with an inventory-management technology, and C items can be controlled with a simple, rules-based system, such as periodic provisioning, which enables planners to source and supply these items with minimum administrative expense.
Multi-criteria-based inventory classification
Castle Metals provides a broad range of metal products, metal processing capabilities and customized supply chain solutions to a variety of industrial sectors. The company has 20 locations throughout North America, Europe and Asia and works with its international original equipment manufacturers to serve their multi-location production requirements and delivery needs. Castle Metals leaders are committed to continuous improvement and empowering their employees to use their expertise and creativity to provide integrated supply chain solutions to customers. This is done by offering timely delivery of high-quality metal products and processing services, and a competitive and sustainable rate of return to its shareholders.
Until recently, Castle Metals used a global inventory classification system based on a single criterion: pound usage. But this did not account for the specific needs of its individual sites, especially those outside of the United States. For instance, an A item in the United States may have little demand in Canada, which would mean that the Canadian location was wasting space storing low-demand SKUs. The supply chain team responded by implementing process improvement that enabled Castle Metals to shift from a global to a local approach to inventory classification and achieve an improved focus on high-demand items at each location.
This modification yielded good results, but there still were too many A items at each location, and they were taking up considerable employee time. So, Castle Metals leaders decided to conduct a pilot for a new, multi-criteria-based inventory classification system. They started in one of the smaller locations, which processes approximately 4 percent of customer orders. This test location mainly deals with four types of metal categories: carbon, alloy, stainless and aluminum. Each has subcategories with different grading specifications, although practically all of them are managed in cylindrical bar shapes.
Structured interviews were conducted with members of the supply chain team, which resulted in the selection of the criteria that would be used for inventory classification at the pilot site. The chosen measures were 12-month sales history, total gross profit, order minimums and sales forecast for the next four months. Supplier lead time also was recognized by all interviewees as an important measure, but it was not used in the pilot. Decision-makers said supplier lead time would be the first standard added when expanding the use of the new methodology.
AHP model implementation
There are several different methodologies for multi-criteria inventory classification, including the analytic hierarchy process (AHP), genetic algorithms, fuzzy logic and neural networks. The AHP framework is commonly used for scoring inventory items and has been proven effective in many situations. AHP is built upon pairwise comparisons of criteria, which Thomas L. Saaty and Luis G. Vargas say in “Models, Methods, Concepts and Applications of the Analytic Hierarchy Process” enable decision-makers to achieve “objectivity through subjectivity.”
Applying the AHP scoring algorithm to these comparisons results in a relative weight for each criterion. After a SKU is rated on each measure, the scores are multiplied by the AHP-derived criteria weights and summed to derive a score for the SKU. The comparisons are made on a scale of 1-9. Figure 1 demonstrates the scheme used to assign a rating to each of the comparisons between X and Y.
Further, if the comparison rating of X compared to Y equals “a” (with “a” being between 1 and 9), then the comparison rating of Y compared to X per AHP rules is:
In the Castle Metals pilot, six comparisons needed to be made across the four specified measures. The supply chain managers did an independent comparison. Then, they were brought together for discussion and further evaluation. Eventually, they reached consensus on the rankings for the comparisons. This resulted in a 4-by-4 matrix of the classifications. Applying the standard AHP scoring method to this matrix resulted in the weights shown in Figure 2.
Determining the scores
The next step was to calculate a final score for each SKU. This was done by taking the SKU’s rating on each criterion and multiplying it by the assigned weight. However, each criterion has its own scale, so the SKU classifications needed to be normalized in order to be comparable across criteria. For each criterion, the maximum and the minimum across all SKUs was calculated. Then, a normalized rating was calculated for each SKU as:
For instance, if the total gross profit varied between $100,000 and $500,000 across SKUs, and a particular SKU had a total gross profit of $160,000, its normalized rating would be:
In this manner, each SKU has a rating between 0 and 1 for each criterion. A higher rating indicated a higher preference for that SKU. Once the ratings were determined, they were multiplied by the criteria weights and summed to give the final score for all of the SKUs at the Castle Metals pilot site.
Next, a Pareto chart was generated based on the scores, which was used to stratify the SKUs. Even when the cutoff for A items was 80 percent of accumulated value, the multi-criteria method resulted in fewer A items compared to the original system. If the cutoff for A was 70 percent of accumulated value, then A would have even fewer items. In the end, the new system refined the classification of 32 percent of the SKUs.
Benefits of stratification
The primary advantages of Castle Metals’ new classification policy include:
- more objective and narrower identifications of A items
- improved allocation of investment dollars
- decreased operational costs
- enhanced customer service levels, especially for A items.
Perhaps most importantly, the process of arriving at the new classification involved the expertise of all members of the supply chain team, as well as marketing managers, which improved buy-in for the inventory management changes. Team members are confident that this initiative will be successfully rolled out at other locations in the future. Furthermore, the Castle Metals leaders look forward to repeating the exercise once or twice each year, enabling fine-tuning and increasing the effectiveness of the company’s inventory classification practices.
Jaime Quilez Calleja is operations area manager for Amazon Spain, where he leads a team of more than 100 associates. He also is a recent master’s graduate in industrial engineering and supply chain management. Calleja may be contacted at email@example.com.
Gurram Gopal, Ph.D., is a professor in the Department of Industrial Technology and Management at the Illinois Institute of Technology. He received a Fulbright Scholar Award to teach and conduct research at the Galway Mayo Institute of Technology. Gopal may be contacted at firstname.lastname@example.org.
Katherine W. Olsen, CSCP, is director of supply chain at Castle Metals. She also is an adjunct professor at the Illinois Institute of Technology. Olsen may be contacted at email@example.com.