Demand Forecasting in Supply Chain Management-(1/2)
Project Leaders — Sundar Raghavan, Anand Modi- SIGMA The Business Club Of NIT Trichy
The article aims to give you a brief overview of demand forecasting in the supply chain domain as well as walk you through the machine learning project by the SIGMA-Analytics team to predict 3 months of sales for 50 different items at 10 different stores through a time series approach.
Demand forecasting in supply chain management refers to the process of planning or predicting the demand of materials to ensure you can deliver the right products and in the right quantities to satisfy customer demand without creating a surplus. Forecast error can result in creating a surplus, which is both wasteful and costly.
The quantitative forecasting method is used when there is existing historical sales data on specific products and a pre-established demand. It requires the use of mathematical formulations and data sets like financial reports, sales, revenue figures, and website analytics. The qualitative method, on the other hand, relies on emerging technologies, pricing and availability changes, product lifecycle, product upgrades and, most importantly the intuition and experience of those planning the forecast.
Forecasting plays three major roles in effective supply chain management:
- Pivotal in strategic planning of Business: Forecasting is the underlying hypothesis for strategic business activities like expansion planning, budgeting, financial planning, risk assessment, and mitigation. Critical business assumptions like turnover, profit margins, cash flow, capital expenditure, etc. are also dependent on Forecasting.
- Initiating all push–processes of Supply Chain: Forecasting is the starting point for all push processes of Supply Chain like raw material planning, purchasing, inbound logistics, and manufacturing. Better forecasts help optimize the inventory levels and capacity utilization.
- Driving all pull–processes of Supply Chain: Forecasting drives all pull-process of Supply Chain like order management, packaging, distribution, and outbound logistics. Better forecast improves the distribution and logistics and increases customer service levels.
Demand Forecasting Techniques
Within the sphere of qualitative and quantitative forecasting, there are several different methods you can use to predict demand:
- Collective Opinion, which leverages the knowledge and experience of a company’s sales team to aggregate historical data on customer demand.
- Customer Survey Method, which provide key information on customer expectations, desires, and needs. This data is useful for creating a sales forecast but is harder to predict actual demand.
- The Barometric Method, which involves using economic indicators to predict trends and measure current, past and future activity.
- The Expert Opinion Method, which involves soliciting expert advice from external contractors to determine future activity.
- The Market Experiment Method, which utilizes market experiments carried out under controlled conditions to inform retailers on consumer behavior.
- The Statistical Method, which allows a company to identify and analyze the relationships between different variables; establish performance history over time, identify trends and extrapolate potential future trends.
How To Forecast Demand
Demand forecasting is valuable to all businesses but is particularly useful to e-commerce brands and retailers, where accurate forecasting can support inventory management efforts and improve the customer experience.
But knowing how to approach something as complex as forecasting accuracy for an e-commerce store is no small task. Fortunately, there are some tried and true strategies that can make the process easier.
Collect the Right Data
For your demand forecast to be successful, you must ensure that you have the right kind of data to make informed business decisions. It’s important to hone-in on the numbers that give you the information you need to make decisions, like pricing trends and how many people visited on your sales channels in a given timeframe.
Try not to focus your data collection efforts on a complete product line. It’s better to concentrate on the products and categories that earn you the most income and are the most popular with customers.
Adjust for Variables
There are many factors that go into the daily interactions that affect sales data. For your demand forecast to be successful, you need to account for any variables that may sway your data one way or another, such as natural disasters or unexpected store closures. Another factor is if the product is seasonal or trendy, as intermittent demand or future demand can make it harder to create an accurate forecast.
Document Sales and Demand Trends
Whichever metric you choose, you’ll need a repeatable data analysis process that accurately depicts whether the forecast is getting better or worse; points to items that need the most improvement; measures accuracy at your procurement lead time and provides accurate information by customer, branch, brand, product and category.
Budget, Purchase, and Allocate Accordingly
Once your demand forecast is in place, the only thing left to do is utilize your collected data to draw up a strategy for how, where and when to allocate your resources and purchasing efforts.
Do read the next article in this series which is a time series machine learning model to forecast the department-wide sales of a retail store and derive appropriate insights from it.