Part art and part science, forecasting how much inventory to order has never been easy.
Inventory forecasting, also known as demand planning, is the supply chain management process of predicting demand for products to ensure they can be delivered and satisfy customers.1 The goal is to strike a balance between having sufficient inventory levels to satisfy customer needs without having a surplus.
The practice of demand planning uses past data, trends, and upcoming events to predict the required inventory levels for a future period to meet customer demand.2 Forecasting also requires a deep dive to identify patterns and trends to adjust inventory to dynamic conditions—without tying up too much cash in excessive inventory or losing sales to stock-outs.
The rise of omnichannel and digital commerce has already shaken up traditional methods of demand planning, sending merchants scrambling to keep up with the demands of selling across multiple channels. These challenges include shorter product lifecycles, lean inventories, expanding “endless aisle” assortment of products (sometimes drop-shipped to the customer by the vendor in the case of online sales and marketplaces), frequent price changes, and promotions.
Demand signals for retail in-store commerce are a different breed than the demand signals for e-commerce and omnichannel. Variables in traditional retail inventory forecasting include price, sales promotion and advertising, assortment, seasonality, and holidays, among others. As a general rule, online sales are less predictable than in-store sales, with greater peaks and valleys. Online sales are more directly correlated to changes in how items are managed and promoted such as flash sales, price-matching, customer reviews, and the availability of competing products.3
The supply chain disruptions of the pandemic only added more variables and uncertainty to the mix of balancing supply against demand to keep the right amount of product in stock at the right place and time.
The methodologies of inventory forecasting have also changed. In the past, companies would rely only on their internal data, such as POS sales data, product availability, supplier information, and promotions data to create demand projections in periods such as three or six months out. Now, a multitude of available external data inputs—such as syndicated POS data, social media, consumer trends, and weather predictions, to name a few—can be combined with internal data to more accurately predict demand patterns in real time.
Much like supply chain management, demand planning is still largely done with a combination of manual, paper-based, and spreadsheet approaches using data from data repositories that are often siloed. About three-quarters of supply-chain functions rely on spreadsheets, and more than one-half of supply chains use older enterprise supply chain planning software such as those by SAP or other providers, according to research by McKinsey & Co.4
Drives to further digitize the supply chain and automate processes are helping to centralize data and improve inventory forecasting and replenishment. Systems such as ERP, inventory management, WMS, and order management, demand planning, among others, all inform the process.
Visibility into data and real-time inventory-related information are key requirements for accurate inventory planning, requiring integration among systems to “talk to each other” and share data in real time. Improved forecasting and analytics tools, enabled by technologies like artificial intelligence (AI), machine learning, and predictive analytics, are equipping companies to create comprehensive models, using them to better forecast demand signals and predict variability in demand.5
With solutions powered by AI and machine learning, merchants can discover specific patterns and predict near-future demands to optimize inventories.6 Armed with these tools, merchants can leverage outside data to build robust algorithms and forecasting models to generate more accurate and reliable demand signals.7
A recent poll of supply chain leaders by McKinsey found that about 20% of companies have already implemented AI and machine learning for some type of supply chain planning activity, and 60% plan to adopt these technologies in the future, with demanding planning point solutions among the type of upcoming implementations.8
Through the use of advanced planning solutions, organizations can not only improve the accuracy of forecast results but optimize replenishment. Several software providers in the supply chain planning space have popped up with solutions that sense demand.
AI-enabled supply chain software company ThroughPut Inc. defines demand sensing as a methodology that combines mathematical techniques like advanced algorithms, AI, and machine learning with real-time information to make a near-perfect future forecast of demand that is accurate based on the current realities of the supply chain.9
Ikea and UPS are among the large enterprises that recently announced plans for some form of demand sensing software that integrates live internal data with external data. U.S. retailer Total Wine is implementing a planning platform with the capability to incorporate external factors such as weather patterns and local events into the demand forecast and run real-time scenarios, allowing the omnichannel retailer to quickly adapt to changing market conditions and best position stock.10
Demand sensing solutions give planners the data to run “what-if” scenarios against any number of variables like timing and length of promotions, and break out expected demand by fulfillment method or other parameters to drive optimal inventory accuracy and replenishment.11
The results? AI makes demand forecasting easier and improves the accuracy of hitting the mark of having a product in stock at the right place and time. Research by McKinsey found that AI-enhanced supply chain management significantly improves inventory forecasting accuracy and optimizes stock replenishment. In some settings, AI-based forecasting can reduce forecasting errors by 30% to 50% and reduce inventories by 20% to 50% in the process, and lost sales from product unavailability may be reduced by up 65%.12
Even with the best tools, human decision-making is still required and that’s where analytics come in to provide actionable insights. Analytics are powerful supply chain and marketing tools for merchants, allowing them to take back some of the control that has tipped in favor of today’s omnichannel consumer that demands always-available product and a frictionless shopping experience.13
Developments in analytics are allowing retailers to put goods in front of the consumer even before they think they want to buy the product. Software provider SAS notes that improved AI-driven predictive or “anticipatory” analytics technology can “accurately anticipate what consumers want or need before they even know it, based not just on past behavior but on real-time information and availability of alternatives that could alter consumer choices.”14
The brands and retailers that adopt AI and machine learning technologies early on as they move toward digitizing their supply chains and update their planning tools stand to come out ahead. In short, a tech-driven approach to inventory forecasting that takes into account more variables with external data inputs, along with robust forecasting tools and analytics, gives merchants the actionable insights they need to improve demand planning.
Copyright Ryder E-commerce by Whiplash 2023. All rights reserved. Use of this website signifies your agreement to the Terms of Use and Privacy Policy.
