Success Strategies for Demand Forecasting in Supply Chain
There’s been a lot of change in how we view supply chain demand forecasting: we moved from a focus on supply—what and how much to supply or replenish—to the demand-driven supply chain, which placed too much emphasis on the intermediate goal of an accurate demand forecast.
To make matters worse, traditional demand forecasting solutions weren’t designed for high variability demand. Inventory mixes and service levels get out of balance across the network—and out of line with business objectives. This leads to excessive costs, waste and obsolescence.
What is demand forecasting in supply chain management?
When planning your demand forecast, it’s important to remember the basics. There are a number of skills needed to do it successfully because there are many moving parts.
Demand forecasting in supply chain management is the process of predicting demand, supply, and pricing for products. It also encompasses competitive intelligence, gathering supplier data and looking for patterns to predict future performance. With today’s supply and demand volatility, demand forecasting needs to account for uncertainty while focusing on the ultimate goal of providing service to your customers.
The top four demand forecasting methods for supply chain success
As the post-COVID supply chain shortage drags on, you might be scratching your head and wondering how to forecast amid all the uncertainty. Whether you’re in manufacturing, retail, or another industry, here are four strategies that supply chains just like yours have used to win at demand forecasting.
1. Plan for uncertainty with probabilistic forecasting
Getting your forecast right is an unrealistic goal, so your focus should be on the uncertainty intrinsic in demand. That’s why probabilistic forecasting is a game changer for supply chains–helping companies improve their resilience during times of crisis, serve their customers more effectively and ultimately provide the highest returns to their shareholders.
At the height of the pandemic, planners scrambled to find ways to better understand volatile and complex demand patterns and that’s where probabilistic forecasting shines because it takes uncertainty into account to help you manage risk.
The image compares a statistical forecast to throwing 2 dice (left) and its probabilistic forecasting equivalent (right).
Image source: Stefan de Kok
2. Improve supply chain forecasting with machine learning
According to our ToolsGroup/CSCMP Digital Transformation in Supply Chain Planning report, machine learning is one of the top three technological investments supply chain leaders plan to make to transform their supply chains. Increasing forecasting complexity and rapidly shifting consumer demand are often exacerbated by seasonality, new product introductions, promotions, and causal factors (e.g. weather, social media), making demand planning extremely complex.
Today, businesses are using machine learning for demand forecasting more than any other area in supply chain planning. As you take steps to improve your supply chain demand forecasting, it’s crucial to understand your forecasting maturity stage and how to take the next step forward. Incorporating machine learning into your demand planning can decrease forecast error and lost sales by over 30%.
3. Begin to automate demand forecasting techniques
The role of the supply chain planner has been elevated in the face of growing supply chain complexity. Supply chain planning automation frees up planners to make strategically important decisions, and also makes a new level of business results possible.
Elements of supply chain planning are ideal tasks to outsource to machines, which can execute analytical tasks and repetitive calculations faster and more accurately than humans. Machine learning’s ability to find patterns (like seasonality and shopping behaviors) in huge data sets and also get smarter over time makes it the perfect complement to human planning efforts. Retailers and merchandisers with unique knowledge of their customers can find particular benefit here.
By automating planning decisions, Shamir Optical was able to run a much larger network of 20 locations each with 35,000 SKUs using the same number of planners (three).
4. Master supply chain forecasting for intermittent demand
Consumers now require more and more different product options, which means more intermittent demand and slow-moving inventory. The key to forecasting intermittent demand is to accurately and reliably model demand both in the “head” (fast-moving items that are easier to forecast) and the “tail” (slow-moving or intermittent demand items).
The benefits of demand forecasting in supply chain management
Modern demand forecasting software enables an advanced planning process–one with benefits that delivers healthy inventory levels, higher productivity, and increased product availability.
Improved customer satisfaction: When customers want to buy your product, they expect it to be in stock. Fulfilling demand is a core element of the consumer experience, one which, when neglected, has a direct, negative impact on your brand value. The need to meet customer demand is even more important when looking at an omnichannel buying experience.
Better allocation of resources: Meeting customer demand and maximizing sales require not only having the right amount of products, you must have them in the right location. This is particularly critical for retailers: 53% of brands offer in-store pickup and 73% of brands offer store-to-door shipping. Proper inventory allocation enables a customer-centric approach that maximizes revenue with the right product in the right location.
Supply chain cost reduction: When using advanced supply chain forecasting systems, it’s common to see 15-30% reduction in inventory while improving product availability and inventory turns. Companies can also reduce overhead due to overtime and expedited freight by more than 50%.
Reduction in product obsolescence: In industries with shelf-life requirements, this typically results in high inventory write-offs as lots age out, or heavy discounting in industries such as fashion. Better demand forecasting can result in 10-30% reduction in obsolescence.
Higher supply chain planning productivity: By automating forecasting decisions, you can see a 40-90% reduction in planner workload.
Increased sales performance: With reliable forecasting, product availability can be as high as 99% and when combined with multi-echelon inventory optimization, you can reduce lost sales by 20-50%
Reduction in lost sales: While it varies by product and sector, the probability you’ll lose a sale as a result of a service failure is 50% in retail. Probabilistic forecasting helps you get your forecast right more often than using single-number forecasts. With this method, you can increase forecast accuracy/reliability by five percentage points and reduce forecast error by 10-20%. An average company can save $3.52M for every one-percent improvement in under-forecasting error.
Why can’t I use spreadsheets for supply chain forecasting?
- Spreadsheets aren’t designed for collaboration: When spreadsheets are being emailed among departments, it’s hard to determine which is the most current version. And it’s usually impossible to know exactly who changed which values, when.
- Spreadsheets are prone to error: One mis-typed number or formula and the house of cards comes crashing down; and the larger the data set—and the number of people providing input—the greater the risk of errors becomes.
- Spreadsheets are designed to do calculations: They give one result based on a set of values and mathematical operations. They aren’t designed for modeling or what-if scenarios, or managing uncertainty and probabilities. Planning is reduced to a “single number” forecast, which rarely matches reality.
- Spreadsheets don’t support automating forecasting: While certain functions within spreadsheets can be “automated” with macros and formulas, most updating and collaboration efforts are manual.
- Spreadsheets aren’t agile enough: When planners have to rely on multiple people updating and passing around multiple spreadsheets, they simply can’t react quickly to changing conditions.
A better way to manage your supply chain data
Spreadsheets are even more difficult to manage when forecasting intermittent demand, particularly when considering the long tail of inventory.
To reliably forecast intermittent demand, you want to master the shape of its demand probability distribution across a wide variety of demand behaviors. Probabilistic forecasting and advanced analytics help create accurate demand and inventory models that support reliable service levels and stock management—without excessive manual intervention from planners.