In an increasingly unpredictable world environment, practices for planning and forecasting inventory levels that worked in the past need to be reviewed. Rules of thumb and techniques that were used only ten years ago cannot match the number-crunching applications that leading-edge companies are employing.
In the 1990s, APS (Advanced Planning and Scheduling) arose as a solution for making better decisions about inventory. But its complexity led to it only gaining traction with a small number of companies. Then in the early 2000s, a new technology emerged, inventory optimization (IO), which could account for variability and multi-level activity in the supply chain and optimize inventory management policies using a statistical approach to manage both demand and supply variability. More recently, artificial intelligence (AI) has been promoted as the way to address the complexities of inventory planning and forecasting.
Both IO and AI can be used to plan how much inventory to hold where, and when to order it. They have become applicable because the large amount of data they require is now created and stored by businesses, thanks to advances in computer technology.
Inventory optimization is the process of maintaining the right amount of inventory required to meet demand, in the right locations, at the same time reducing inventory-related costs, and avoiding common inventory issues such as stockouts, overstocking, and backorders.
IO can help to address the challenges of managing the three forms of inventory:
- basic stock — what is required to satisfy a demand forecast,
- seasonal stock — how much should be stored in anticipation of demand increases that occur at certain times in the year,
- safety stock — how much should be kept aside as a buffer against uncertainty.
Optimizing the levels of these inventory groups is a key objective. IO takes into consideration storage capabilities, current inventory levels, supplier lead times and schedules, and future campaigns. It uses historical data and applies statistical techniques to allocate resources in the most effective way to satisfy competing requirements.
The steps to undertake inventory optimization are:
- Analyze stock codes for importance and behavior and then classify them into categories;
- Generate the best possible estimate of demand, forecasting for each stock code;
- Model a set of stock policies to determine the optimum balance between customer service and inventory investment to meet the expected demand;
- Replenish stock timeously according to the forecast and the stock policy.
Benefits that have been gained from using IO include:
- improved cash flow,
- optimized warehousing capacity,
- reduced storage costs,
- identification of which items are slow-moving or fast-selling,
- better information to plan in which warehouse to store items based on demand;
- making sure that orders can be fulfilled quickly.
Using AI in inventory management
AI is the term for computer software that processes vast quantities of data to find patterns and make predictions and recommendations based on objectives that are set. AI can perform tasks with a much higher degree of accuracy and speed than humans, and so can provide insights to help humans make significantly better decisions.
AI in inventory management can help companies eliminate time-consuming and tedious tasks. For demand forecasting and planning, AI algorithms can handle a huge number of variables and analyze complex relationships to develop demand plans and predictions that usually exceed human-based forecasts in quality and quantity. The McKinsey report Smartening up with Artificial Intelligence showed that AI was able to reduce:
- forecasting errors by 30 – 50%;
- lost sales by up to 65%;
- costs related to transport and warehousing by 5 to 10%;
- supply chain administration by 25 to 40%;
- inventory by 20 – 50%.
Manufacturers can use AI to manage supplier quality and performance, discovering who their best and worst suppliers are, and which inventory reception areas are most accurate in catching errors.
The Economist has mentioned the role of AI in inventory management and demand forecasting, noting an example where AI can forecast in what order items will arrive at and leave a warehouse, so that pallets can be put in the right position.
Challenges of IO and AI
Key technical issues to address when using IO or AI for supply chain excellence are data accuracy and timeliness — others are skills and culture.
Data-related problems are less likely to occur if the IO or AI application is integrated with an ERP system. An ERP solution eliminates issues of data accuracy and timeliness, and enables data to be easily synchronized with the application. These issues have to be separately and intensively addressed when different point solutions are used.
To get the most out of AI requires organizational changes. New skills are needed to manage and use the technology. Furthermore, a culture of data literacy should be developed to help staff know how to make appropriate data-driven decisions.
The ability to manage inventory better
Inventory management is not just about shipping goods to customers. It’s about having the inventory in place before customers order it, which requires extremely precise forecasting, based on the analysis of large amounts of data. Using an IO or AI solution as part of an overall ERP system, can allow companies operate more efficiently, improve their services, attract more customers and by reducing costs, offer lower prices.