If manufacturers want to generate higher levels of profitability, inventory management needs to be better understood.
Manufacturers often manage dozens of major product lines with variations that require tens of thousands of parts from hundreds of supplier manufacturers spread across the globe. Holding the associated stock to accommodate each product configuration quickly becomes one of a company’s most expensive ongoing investments and can tie up to half of its available capital.
The mandate to optimize is further magnified during times of disruption when historical forecasts tend to produce large errors both on the demand and supply side. Deficient or excessive inventory levels directly correlate with cost and risk, and yet few companies have successfully implemented the necessary enterprise AI technologies to optimize their inventory, at scale.
The manufacturing research field is flooded with doctorate theories detailing rules-based inventory optimization policies. In parallel, we have seen the proliferation of AI, machine learning, and elastic cloud computing technologies that have enabled manufacturers to gain a near real-time view into their supply chain.
Companies that have embraced these new capabilities are dedicating entire teams to understanding how to use AI and ML to reduce inventory while maintaining service levels. The results of AI-based methods are far surpassing the static predictions generated by a rigid rules-based approach, setting a new and higher bar for the industry.
Manufacturers that have deployed stochastic optimization programs attribute up to millions of dollars in annual savings to decreased inventory carrying costs and operational efficiencies. These companies have been able to aggregate their disparate supply chain data into a unified view, apply machine learning to understand variability factors in their inventory, generate risk scores that inform business decisions, and preemptively take data-driven action to ensure customer needs are met.
Before a manufacturing organization can begin to optimize its inventory management processes, it is critical that its leadership team is clear on the following three elements:
- The tangible and financial benefits that AI-based inventory optimization will bring to supply chain management
- The organizational benefits that inventory insights will provide, as well as the reconciliation burdens they will alleviate
- The concrete steps required to successfully implement and scale new technology within the organization
How AI-based inventory optimization can create value
Let’s look at some of the key areas where manufacturers can apply machine learning and artificial intelligence to the supply chain to dramatically alter the assumed cost and risk of carrying excess inventory.
|Inventory reduction||They can invest in capital previously tied up in excess inventory against profitable projects.|
|Taxes||There is no tax advantage to keep an inventory that is larger than necessary for your business purposes. As optimization allows fewer parts to be procured from suppliers, sales taxes are reduced.|
|Insurance||Insurance expenses are based on the amount of materials managed/handled as well as the initial cost and potential value of the finished goods; as optimization enables a reduction in overall inventory, insurance expenses should drop as well.|
|Shipping costs||Inventory is reduced as fewer parts are ordered, which means manufacturers incur lower shipping costs, particularly when those shipping costs are based on quantity of parts. Additionally, optimized inventory drastically reduces their need to use expedited shipping, which typically costs 3 – 10 times the standard rate.|
|Productivity improvement||Productivity improvements accrue to buyers, planners, analysts, and material specialists.|
|Storage costs||Reduced inventory levels provides manufacturers with the opportunity to reduce square footage in their warehouse and distribution centers. In addition, this frees up space for them to store other products and may result in additional benefits, such as reduced energy costs for heating or cooling.|
|Material handling costs||Many warehouses and distribution centers are staffed with temporary workers. As fewer materials need to be handled, manufacturer labor costs for loading, unloading, and moving materials should go down. At warehouses staffed by permanent workers, hours are freed up that they can assign to value-add activities.|
Technology fueled clarity into business processes
Along with the material planning and customer service impact, insight into optimal inventory levels empowers all areas of a manufacturing organization to better understand its connection to inventory value. This objective transparency mitigates costly inventory issues, particularly those associated with E&O – excess and obsolete – stock. Other inventory issues include:
- Not applying the true cost of inventory: Function and SKU level insights alone generally do not provide sufficient accountability for manufacturers to determine the cost of inventory. Other aspects of the organization, such as sales, manufacturing, or purchasing/suppliers can have a direct impact on inventory levels if they lack current inventory insight. This may have a dramatic effect on the sales & operations planning (S&OP) function, resulting in inaccurate forecasts, missed sales opportunities, and added risk to customer commitments. Manufacturers with a clean line of sight into all inventory, its location, ownership, life cycle, and value helps them to make informed decisions.
- Poor end of life planning: Many manufacturers overlook the need to review their contractual commitments for holding inventory. Holding onto inventory that is slow moving or obsolete – or worse, ordering more of it – consumes scarce resources and prevents the application of higher value outcomes. These hidden costs are rarely factored into the sales account management cycle.
- Underestimating the financial impact: E&O inventory problems are often misunderstood and understated by manufacturers. It is not uncommon to learn that much of a company’s material is obsolete, not fully well-known or – in some extreme cases – that warehouses contain material that is several years old and may not even be recorded on company records. Associating a price tag with the inventory provides a more tangible and realistic understanding of the problem. And, applying financial metrics to the issue usually drives action and improvement.
Proven methodology to implement and scale inventory optimization programs
To implement these technologies successfully, manufacturing companies need to apply core concepts that are proven in production. These initiatives require support from senior leadership as well as complete alignment between supply chain, finance, and IT departments to roll out and attribute value to the initiatives.
Companies that build and attribute value to their inventory optimization programs generally follow these three key steps:
- Identify high-value use cases that will return results within 6 to 12 months: Organizations are dynamic and can transform rapidly. Program initiatives frequently become burdened with complex deliverables that are not clearly defined. It is essential for operational teams to demonstrate company-certified ROI, not only to yield positive results but also to expand usage and advocacy within the company.
- Leverage off-the-shelf SaaS solutions when possible: Many IT organizations underestimate the complexity of building sophisticated AI-based applications that can scale beyond the first use case. Today, there are commercial AI solutions available and in production. They don’t need to be manually built using cloud service provider primitives alone which can introduce high risk and cost. These modern solutions are tried, tested, and proven at scale. They enable data scientists and developers to focus on deriving real value. For example, some of the most sophisticated manufacturers in the world have successfully transformed their inventory management processes by using pre-built applications such as C3.ai™ Inventory Optimization.
- Create a transformative culture of innovation: Manufacturers that are successfully deploying AI and IoT applications, rapidly and at scale, recognize that they need to dedicate time and resources to re-tooling their teams for these initiatives to succeed. Global leaders such as Baker Hughes, Shell, ENGIE, and others are developing Centers of Excellence to train their developers and data scientists and empower them to build the next generation of applications for their organizations.
The future of inventory optimization
Inventory forecasts can’t be permanently “fixed” – they will always be susceptible to real-time changes and, therefore, imperfect by definition. During a period of disruption, as we are currently experiencing, effectively managing inventory can be even more challenging.
Historical time series-based models using data from year over year time horizons are unable to accurately inform predictions and the public and fiscal impacts of ill-informed planning can be magnified. Instead, a stochastic optimization approach provides inventory planners with the ability to adjust inventory forecasts as new data is ingested and processed daily. Applications such as C3.ai Inventory Optimization are emerging as valuable tools that can deliver new insights into the operation and strategy for inventory management and running a manufacturing business.
As the manufacturing industry evolves, we are seeing leading manufacturers embrace the internal (and external) mandate to digitally transform their supply chain beginning with inventory optimization management. Notable Fortune 100 organizations and other capable incumbents are readying themselves for the future by gaining complete insights into their inventory and leveraging advanced AI solutions to optimize it.
Dan Barrett is a 25-year software veteran focused on the application of AI and ML to supply chain operations, sales & operations planning, inventory, warehousing and transportation management. You can reach Dan at email@example.com to further discuss this blog and inventory optimization use cases.