What is a safety stock?
The durability of a business whose activity is based on the sale of physical products depends greatly on its ability to meet customer demand and avoid sales losses. Out-of-stock situations are part of the events that can alter the functioning and the reputation of an electronic or physical business. They can emerge when there are malfunctions in the supply chain. It is therefore important to prepare for this potential risk by implementing a sales forecast that takes into account safety stock.
How to predict, or even anticipate, stock-outs? What criteria should be taken into account when developing safety stocks? What tools can be used to optimize out-of-stock forecasts? We'll review here all the dimensions to be taken into account in the management of buffer stocks for your physical or digital points of sale.
Causes of stock-outsInventory forecasting at your various points of sale, whether physical or digital, aims to meet customer demand and avoid lost sales in the event of stock shortages. The causes behind stock-out situations affect different levels of the supply chain. For example, stock-outs can be linked to extended delivery times due to supplier delays. They can also occur further up the supply chain in the event of a lack of raw materials required to create the products sold by the company concerned. Finally, factors external to the supply chain can disrupt its operation and have an impact on your inventory levels, such as carrier strikes, weather conditions or large-scale health issues.
Among the reasons closely related to the internal functioning of an organization, we can mention:
- poor forecasting or errors in estimating orders;
- the decision to stop marketing a given product or item;
- an error in the accounting of physical products (error due to manual calculations).
Finally, stock-outs can occur when a product is successful in ways not anticipated by the company's operational teams. The high demand induced at that time implies an inability to meet customer needs.
The different storage costs
Companies whose activities are centered on the sale of goods, such as mass distribution companies, retailers, pharmaceutical product specialists, e-tailers, etc., must manage the flow of products and raw materials by taking into account various constraints. Indeed, inventory forecasting requires the consideration of customer needs satisfaction as well as the reduction of various logistic costs. Among the costs of replenishment of goods, we find those of transport and storage.
Within the costs incurred by inventory management, we can distinguish between the costs of fixed installations. These are all expenses related to the acquisition of premises dedicated to the storage of products. The costs spent on setting up shelving, maintaining the physical premises (insulation, renovations, repairs), rental costs (if the company does not own the storage location), insurance costs for the premises, taxes and duties related to the management of real estate, depending on the size of the inventory, can be grouped together under facilities and storage costs.
The immobilization of goods generates inventory carrying costs. These can be broken down into cash tied up in inventory, electricity or heating costs, labor costs dedicated to managing the inventory, risk costs invested to protect the goods from theft or damage, etc.
The differences between safety stocks and minimum stocksThe safety stock responds to customer demand and ensures the continuity of sales when the company encounters complications during the replenishment phase. Called safety stock in English, it is a quota of goods whose quantity depends on the hazards faced by the business in question, as well as the storage costs. The higher the contingencies and the lower the storage costs, the higher the safety stock.
Among the hazards that a business may face are: delivery delays, stock-outs on the side of suppliers and their suppliers, incidents or external events affecting the supply chain such as carrier strikes, etc. Another factor determines the size of this stock: the quality of customer service. Indeed, in order to guarantee customer satisfaction and minimum response and delivery times, a given organization may decide to increase the level of safety stock.
The minimum stock level is calculated according to the customer demand. The aim behind the establishment of these levels is to avoid having to resort to safety stock (except in the event of a hazard). This stock, known as cover stock, ensures the continuation of sales during the delivery period of an order. A third level of stock links the two previous ones: it is the alert stock. It is equivalent to the sum of the minimum and safety stock levels. Once reached, the goods orders are triggered.
The criteria for setting up a safety stock
The calculation of safety stock levels is based on data regarding delivery times and average weekly or monthly sales. Moreover, not all the references sold within your business require the implementation of a buffer stock. Therefore, it is necessary to establish the list of products to be included in the safety stock calculations. The criteria for selecting these items include customer demand (high or low) and the level of customer service you wish to achieve. Indeed, if the speed of delivery or the continuous availability of certain products at the point of sale is one of your priorities, the buffer stock dedicated to them will be all the more important.
In addition, the choice of safety stock levels must take into account the management and storage costs of the items in question. In addition, in the case of food and perishable products, inventory managers also manage the data related to the time dimension in their calculations related to the level of safety stock to be maintained. Finally, the effects of fashion and seasonality can affect certain product references. These may therefore require the setting up of safety stocks according to vacations, vacations or other specific events.
Optimizing safety stock management with machine learning
rands and companies whose activity is based on the management of a large volume of products, such as mass distributors, retailers and e-tailers, are faced with an increase in the complexity of their logistics process (large amount of data). Thus, generating reliable predictions concerning the safety stocks to be expected for each stock management unit (product reference or SKU) according to each point of sale cannot be done without the use of tools capable of managing large quantities of data.
Indeed, on a human scale, the taking into account of the various factors influencing the evolution of stocks cannot take place. And when it is carried out on tools such as Excel or certain ERPs, it is only done periodically. All of these constraints contribute to the inaccuracy of the predictions generated and imply increased storage costs and a lack of reactivity to customer demand.
Artificial intelligence now offers models capable of capturing the different dimensions to be processed as input in order to ensure personalized forecasts for each item according to the sales unit where it is marketed. Indeed, the computational power and accuracy of machine learning tools offer the possibility to take into account both the classical sales forecasting requirements, but also to efficiently calculate the necessary safety stock levels within a particular sales unit.
The granularity of the definition of these buffer stocks can characterize each SKU for each store. By using AI-driven predictions, a company can move away from generic forecasts for all its stores. It also puts an end to manual or statistical safety stock forecasting processes that are tedious and unable to take into account changing trends within individual retail spaces.
These data science tools can also cross-reference exogenous information (vacation schedules, public holidays, weather data, supplier and supply chain data, etc.) with your company's historical data. The inventory predictions generated by these algorithms are more holistic than those generated by manual processing. Finally, predictive tools based on machine learning offer a level of temporal accuracy by processing historical and real time data streams. Inventory changes are then analyzed almost instantaneously.