Knowledge base for supply chain optimization

Knowledge base

Supply chain topics


Forecasting topics


  • ABC analysis

    ABC analysis is an inventory categorization method which consists in dividing items into three categories, from the most valuable items to the less valuable. This allows to give them a weighed treatment to better optimize the inventory.
  • Backorders

    Backorders represent purchase orders made to the supplier for products that are already out of stock from a given location being served. Backordering is the process of selling inventory the company doesn’t have on hand.
  • Container shipments

    Container shipments represent a specific supply constraint frequently associated with oversea import. This constraint must be satisfied with preserving the demand forecasts that are established separately.
  • Economic order quantity (EOQ)

    EOQ is the purchase order quantity for replenishment that minimizes total inventory costs. The problem lies in getting the EOQ formula right, since the classical Wilson formula is a poor fit for modern supply chain.
  • FIFO inventory method

    Method stating that the first goods purchased are also the first goods sold. FIFO inventory can be considered as a theoretical model of the actual flow of goods. It can also be considered as a supply chain practice, intended to limit expiration issues that negatively impact stocked goods.
  • Fill Rate

    The fill rate is the fraction of customer demand that is met through immediate stock availability, without backorders or lost sales. The fill rate appeals to practitioners because it represents the fraction of the demand that is likely to be recovered if the inventory performance was to be improved.
  • Financial impact of accuracy

    More accurate demand forecasts are obviously a good thing. However, the quantitative assessment of the financial gains generated remains a fuzzy area for many retailers and manufacturers. This article details how to compute the benefits generated by an improved forecast.
  • Inventory accuracy

    The notion of inventory accuracy refers to all the discrepancies that exist between electronic records that represent the inventory and the physical state of the inventory. This article describes a simple way to measure it with the MAE, as well as the extent of the problem in retail.
  • Inventory control

    Definition and insights. Inventory control covers diverse aspects, including the management of inventory (recording both quantities and locations of items), but also the optimization of the inventory (carrying costs, stock-out costs, ...).
  • Inventory costs (carrying costs)

    Inventory costs are the costs related to storing and maintaining its inventory over a certain period of time (ordering costs, carrying costs with capital costs, storage space costs and so on). This article categorizes, defines an estimates those costs.
  • Inventory turnover

    The inventory turnover is the number of times the inventory must be replaced during a given period of time, typically a year. It reflects the overall efficiency of the supply chain. This articles provides a definition of this notion, as well as formulas and insights.
  • Lead time

    The lead time is the amount of time between the placing of an order and the renewed availability, after the receipt, of the goods ordered. It has a direct impact on the inventory needed and is a key factor in forecasting. However, the lead time as it is required by our forecasting engine has a slightly different definition.
  • Lead demand

    The lead demand (also called lead time demand) is the total demand between now and the anticipated delivery time, if a backorder is made now to replenish the inventory. Its accurate estimation is critical to assess the minimal amount of inventory needed to reach specific service level objectives.
  • Min/Max Planning

    The Min/Max inventory ordering method is a basic reordering mechanism that has been implemented in many ERPs and other inventory management software. The “Min” value represents the stock level that triggers the reorder and the “Max” value represents the new stock level targeted after the reorder.
  • Minimal Order Quantities (MOQ)

    A MOQ indicates that a supplier won't accept a purchase order below a specified threshold. The general MOQ problem consists of computing the (near) optimal purchase orders that both satisfy all the MOQ constraints while maximizing the economic returns associated to the units purchased.
  • Multichannel Order Management

    Multichannel Order Management solutions (MOMs) are B2B software intended for merchants with a primary focus on online commerce. MOMs address two key concerns. First, they consolidate orders across all channels into the same physical inventory. Second, when prices are changed, they dispatch the new prices across all channels.
  • Perpetual Inventory

    Perpetual inventory refers to a system, a software in practice, where the information on the inventory quantity and availability of the SKU (stock keeping unit) is updated continuously based on the records associated to business operations. The goal of perpetual inventory is to maintain an accurate estimation of the real stock levels with as few stock counts as possible.
  • Phantom Inventory

    Phantom inventory refers to goods that are recorded as available on-hand at a storage location within an inventory management software, but that are not actually present. This article describes its root causes and impact on inventory availability.
  • Prioritized Ordering

    Prioritized ordering policy emphasizes multi-item decisions, where each item competes for capital allocation with all the other items. In practice, prioritized ordering provides much more fine-grained control over inventory performance, and, when the proper predictive technologies are available, prioritized ordering allows to reach superior inventory performance.
  • Product life-cycle

    The life-cycle represents the various market stages - development, introduction, growth, maturity and decline - that occur for the vast majority of consumer packaged goods. From the inventory management viewpoint, it is one of the major demand patterns, along with seasonality or trend, that needs to be accounted for.
  • Reorder point

    The reorder point (also known as ROP, reorder level, ...) is the inventory level of an item which signals the need for a replenishment order. It is classically viewed as the sum of the lead demand plus the safety stock. At a more fundamental level, the reorder point is a quantile forecast of the future demand.
  • Replenishment

    The (inventory) replenishment is an operation that consists in making the stock full again in order to avoid stock-out. Replenishment is typically initiated by a backorder passed to a supplier or to a manufacturer, possibly sent through EDI.
  • Safety stock

    This guide explains how to optimize inventory levels, in retail and manufacturin,g by adjusting safety stocks to their optimal level. The theory is illustrated with Microsoft Excel. Advanced notes are available for software developers who would like to reproduce the theory into a custom application.
  • Service level

    The service level is a percentage that represents the expected probability of not hitting a stock-out. It represents a trade-off between the cost of inventory and the cost of stock-outs. In this article, we detail diminishing returns involved when shifting toward higher service levels.
  • Stock-keeping unit (SKU)

    In the field of inventory management, SKU refers to a specific item stored in a specific location. It is intended as the most disaggregated level when dealing with inventory. This articles provides definition and insights on the notions of SKU and stock on hand.
  • Backtesting

    The notion of backtesting refers to the process of assessing the accuracy of a forecasting method using existing historical data. This articles explains this process and the mistakes to avoid.
  • Continuous Ranked Probability Score (CRPS)

    Probabilistic forecasts assign a probability to every possible future. Yet, all probabilistic forecasts are not equally accurate, and metrics are needed to assess the respective accuracy of distinct probabilistic forecasts. CRPS is an accuracy metric dedicated to probabilistic forecasts.
  • Forecasting accuracy

    The accuracy, when computed, provides a quantitative estimate of the expected quality of the forecasts that can serve several purposes. This articles defines the different metrics used to assess this accuracy and provides a few warnings against misconception.
  • Forecasting methods

    This guide explains elementary forecasting methods to anticipate customer demand that can be readily applied into Microsoft Excel spreadsheets. Advanced notes are available for software developers who would like to reproduce the theory into a custom application.
  • Obfuscation

    Time-series data are sometimes highly sensitive. Obfuscation methods can be used to protect it. This article describes how simply obfuscation can be performed on time-series data.
  • Overfitting

    This links to one of our blog articles and a video about overfitting, a problem with a strong impact on forecasts that has been puzzling mathematicians since the 19th century. Basically, the notion of overfitting refers to a modeling error that occurs when a statistical model is too closely fit to a limited set of data.
  • Pinball loss function

    This article provides the definition and formula of the pinball loss function. This function is a way of evaluating the accuracy of a quantile forecasting model, that cannot be evaluated with the same metrics as classic forecasts.
  • Probabilistic forecasting

    A probabilistic forecast represents an estimation of the respective probabilities for all the possible future outcomes of a random variable. In contrast to single-valued forecasts, such as median time-series forecasts or quantile forecasts, the probability forecast represents a probability density function.
  • Quantile regression

    Quantiles are particularly useful for inventory optimization as a direct method to compute the reorder point. The quantile regression introduces on purpose a bias in the results. Instead of seeking the mean of the variable to be predicted, it seeks the median and any other quantiles (or percentiles).
  • Seasonality

    Seasonality is one of the most frequently used statistical patterns to improve the accuracy of demand forecasts. It refers to the fact that sales (and underlying time-series) undergoe a predictable cyclic variation depending on the time within the year.
  • Time-series

    A time-series is a list of dates, each date being a associated to a value. Time-series are a structured way to represent data. In retail or manufacturing, time-series are important because they are the most canonical representations for the flow of goods either sold or produced.

Antipatterns topics


  • Supply Chain Antipatterns

    Antipatterns aren’t just bad practices. Antipatterns are common responses regarded as good practices, which unfortunately yield serious unintended consequences and defeat the expected benefits.
  • Devil's advocate

    Management has met with an enterprise vendor, and everybody was immediately seduced by the vendor’s proposed solution. The vendor was promising an easy path: all the complexity would be handled by the vendor himself thanks to their unique technology and methodology.
  • The Non-Euclidian Horror

    Every time a new challenge arises, IT systems are adjusted at a minimal level to deliver the expected results. Narrow incremental changes are now strongly favored over any substantial evolutions. Nobody really understand any more what is going on within the company systems.
  • The 100% service level

    Achieving a 100% service level, in order words, 0% of stock-outs or delays, is a fantasy that is unfortunately still chased by too many companies.
  • The Jedi initiation

    Top level executives do not have the time to be hands-on with operational projects, and neither do they have the time to figure out the ins and outs of the myriad of new technologies that could improve every aspect of their supply chain.

Pricing topics


  • Bundle Pricing

    With bundle pricing, the seller lowers the variance on this willingness to pay and increases its profit by selling bundles of products instead of selling all products separately. Moreover, the seller makes bundles in order to fulfill a specific need of its customers.
  • Competitive Pricing

    Competitive pricing consists of setting the price at the same level as one’s competitors. This method relies on the idea that competitors have already thoroughly worked on their pricing. In any market, many firms sell the same or very similar products, and according to classical economics, the price for these products should, in theory, already be at an equilibrium (or at least at a local equilibrium).
  • Cost-Plus Pricing

    Cost-plus pricing consists of setting the price based on the production cost and the desired level of mark-up. This method allows a company to secure margin and is easy to compute on a large amount of products. This method is widely used by retail companies today on at least some of their products.
  • Decoy Pricing

    Decoy pricing is a pricing method that is meant to “force” customer choice. When customers make a purchase they must often choose between products with different prices and attributes.
  • Long-term maintenance agreement pricing

    When a company commissions expensive industrial equipment, the financial risk of maintenance can fully or partially transferred to the vendor through a long-term maintenance agreement. In order to ensure a profitable contract for the vendor, an accurate assessment of the part servicing costs is required.
  • Long-term pricing strategies

    Prices represent a trade-off between short-term and long-term strategies, profitability and market share as well as levels of cash flow. As a result, a company sets its prices in order to fulfil its different objectives such as profitability, cash flow or growth.
  • Odd Pricing

    Odd pricing is a pricing method aimed at maximizing profit by making micro-adjustments in pricing structure. It relies on the assumption that consumers are calculation-averse and will therefore only read the first digits of a price when making their purchasing decision.
  • Penetration Pricing

    Penetration pricing is a very aggressive type of pricing. When using this method, a firm first sets its prices at a very low level (sometimes even with negative margin) in order to increase customer demand. After this, the company raises the price again, hoping to capture the same level of customer demand as with its previous very low level of pricing.
  • Price Elasticity of Demand

    In modern microeconomics, price elasticity measures the correlation between the variation in demand and the variation in price. If the market is elastic, a tiny change in price results in a large change in sales volume. If the market is inelastic, in this case, a large change is price results in a tiny change in sales volume.
  • Price Skimming

    Price skimming can be considered as a form of price discrimination. On the release of a new product, a very high price is set at first in order to maximize profit by selling the good to “early adopters”. The price slowly decreases with time in order to maximize profit by selling the product to other types of customers.
  • Repricing software (Repricer)

    In commerce, a repricing software - or repricer - is a solution to automatically recompute prices of all the items being sold depending on the market conditions. Repricers typically rely heavily on competitive intelligence tools that collect the prices of the competition.
  • Styling prices for retail

    The visual aspect of prices is essential in the retail industry. Better price visuals can significantly impact the demand and therefore the profit too. In order to maximize profit, the price should be displayed in small size, on the bottom left of the label for offline retail.