We often say that the supply chain is either order-driven (pull) or forecast-driven (push). In fact, whether it is order-driven or forecast-driven, from the perspective of the whole supply chain, it is ultimately forecast-driven, because one person's order is destined to be another person's forecast. For example, when a female user buys clothes online, she must place an order with the merchant on the platform, based on her prediction of wearing the clothes in the future.
Demand forecasting is the original driving force of supply chain.
The goal of demand forecasting is "to be as accurate as possible and correct the deviation as soon as possible". This needs to solve three problems: first, how to forecast the demand; Second, who will make demand forecast in order to effectively connect sales and operations; Third, if the forecast is wrong, how to establish a rolling planning mechanism, find it as soon as possible, correct and remedy it as soon as possible?
Demand forecasting and inventory planning is actually a game around forecasting risks. A good demand forecast needs to be adjusted regularly, but this does not mean that it can be adjusted at will. The flexibility of the supply chain is not infinite. When entering a certain time window, the adjustment of demand forecast should be controlled to protect the efficiency of supply chain.
Otherwise, it will cause high operating costs and waste of production capacity. Frequent adjustments will disrupt the overall production and distribution arrangements and make the overall delivery more unpredictable. The more unpredictable, the more human intervention is needed, which will fall into a vicious circle, increase uncertainty, and eventually translate into cost and inventory.
Some futures brands habitually pressure franchisees, forcing franchisees and stores to place orders several months in advance. This order is actually a demand forecast. If the forecast accuracy is low, it will cause inventory backlog.
The ordering goal of a general ordering meeting is to "start with (historical) data and judge the end according to historical results", which is produced on the basis of analyzing historical sales data and consulting the feedback from sales departments. But the most important step, that is, reaching an agreement with franchisees, is often impossible.
The clothing fair is a typical example.
At present, some clothing brands have begun to make supply and demand forecasts, predict regional and national demand, and the time granularity is more detailed. There are special departments to understand market trends, consumer preferences and competing products information. Make an overall forecast, control the total amount and produce an appropriate number of products. Because the time granularity is finer, predict the delivery plan in the next one to two weeks. Even if the prediction is wrong, the impact is limited and it is easy to correct.
The problem of how to forecast has not been solved, the accuracy of demand forecasting is low, the forecasting risk is high, and no one is willing to take risks, so demand forecasting has become a major focus of the game between enterprises and functions; The result of the game, on the one hand, makes the wrong person make predictions, on the other hand, it also encourages information asymmetry, which is not conducive to the improvement of prediction accuracy.
What is the demand forecast?
Let's discuss what demand is first. Demand is the quantity of goods and services that customers can buy in the company assuming sufficient production capacity and no other restrictions. Then demand forecasting is a company's accurate measurement of future demand under a series of assumptions.
What are the hypothetical conditions? It contains the company's internal assumptions and external assumptions. Internal assumptions mainly refer to some company activities that stimulate demand changes, such as advertising, publicity activities, increasing distribution channels, and adjusting commodity prices. External assumptions mainly predict the future economic level, such as industry market, international or national events, bank interest rates, material inflation, competitor trends, etc.
And calculation is a kind of speculation about the future. So it is not accurate. Generally speaking, the accuracy of prediction is only between 50% and 60%, which means that all predictions are wrong. All our work is to use the existing known conditions as much as possible to make it as accurate as possible, and the mistakes are not so outrageous.
Then, since all the forecasts are wrong, why do we need to make demand forecasts?
First of all, demand forecast is a theoretical conclusion based on historical data and future forecast, which is helpful for managers to make decision-making reference for future sales and operation plans, targets and capital budgets. Secondly, demand forecasting can play a recommendation role in purchasing plan and warehouse operation resource allocation, which is helpful for purchasing department to make purchasing plan, and warehouse to make production scheduling plan in advance to reduce the influence of business fluctuation.
If there is no demand forecast, many decisions about sales, procurement and financial budget within the company can only be made by experience, which will lead to insufficient market forecast and problems such as inventory, capital backlog or shortage.
Of course, although demand forecasting is very important, it cannot be used as demand planning or sales target.
The so-called forecast is a hypothesis about the possible future situation, which is essentially a speculation and can only be used as a reference, while the demand plan is a decision made to ensure that the goal can be achieved, which is more authoritative and executable, and the sales goal is the desired result. These three cannot be confused.
To forecast demand, we must first understand the five basic dimensions of forecasting:
Forecast dimension. The predicted granularity is by package or box, by item or by brand. Generally, the finer the granularity, the more variables and the lower the prediction accuracy.
Forecast span. The current forecast is the future demand data, such as the next two months or six months. Generally speaking, the greater the span of the prediction interval, the lower the prediction accuracy.
Forecast interval. How often the demand forecast is updated, such as once a month or once a week.
Forecast unit. Physical measurement standards of demand forecast data, such as parts, parts, yuan, grams, etc.
Forecasting mechanism. Describe the relationship between the granularity of different dimensions and the relationship between different dimensions.
After understanding the basic attributes of demand forecasting, we have the basic skills of forecasting demand. Demand forecasting methods are mainly divided into quantitative forecasting and qualitative forecasting.
quantitative forecast
The method of quantitative forecasting is to explore the law of demand through the analysis of historical data, and the popular explanation is to find the law according to historical data. There are mainly two foreseeable demand patterns:
The first is the time-based demand model. This pattern can be recognized because it will recur predictably at some time. For example, the demand for cruise ships in spring and summer is higher than that in winter, and the sales of roses and chocolates are better every Valentine's Day. In order to find and predict this demand pattern, it is suggested to use time series statistics.
The second is other factors that can affect demand besides time. This model is recognized because some quantifiable variables have a predictable impact on demand. For example, the products of Coca-Cola Company are very sensitive to promotion, and if they are discounted, customers will buy more. For this model, the best solution is regression analysis.
qualitative forecasting
Qualitative prediction method, also called subjective prediction method or judgment prediction method, is a method of collecting and sorting out the opinions, knowledge and intuition of experienced people and transforming them into prediction results.
Under what circumstances is qualitative analysis suitable? There are three main types:
There is no historical data for reference when the new product goes on the market.
The emergence of some new situations has changed the existing demand pattern. For example, because of the epidemic situation, policies and other reasons, the sales of some products have been seriously affected, and we can't just look at historical data through quantitative analysis.
Products with little correlation between historical demand data and future forecast. For example, project-based or customized products.
Enterprises want to do a good job in demand forecasting, in addition to supplementing market information, there are two key factors:
Enterprises should provide a performance evaluation mechanism for forecasting, pay attention to forecasting, distinguish rewards and punishments, diagnose performance evaluation problems in time, and encourage continuous improvement.
No customer will buy more products because of good forecast, but demand forecast can be transformed into high-quality business decisions, thus improving inventory turnover, service quality and reducing supply costs. These improvements will encourage users to buy more products of the company.
The error rate can be used to evaluate the performance, identify the deviation of demand forecast and evaluate the accuracy. The calculation formula is: error rate = (forecast demand-actual demand) * 100%/ actual demand.
Realize the integration of supply and demand of enterprises. Excellent forecasting = forecasting algorithm and modeling+supply and demand integration. The so-called integration of supply and demand refers to mobilizing the production department, logistics department and purchasing department responsible for supply, as well as the sales department, financial department and company executives responsible for demand to jointly formulate a unified and far-sighted plan and make decisions on the optimal and balanced resources for a series of organizational goals.
The efficient integration of supply and demand is reflected in three aspects:
Cultural aspect: openness, transparency, unity and cooperation and common goals are needed.
Process: All links are closely connected, unimpeded and visible throughout.
Tools: Use the right system to convey the right information to the right people.
Because prediction is a guess about the future, it is always wrong. It is not difficult to predict in itself. The difficulty is how to minimize the degree of error.
Forecasting is not a department's business. Sales and marketing departments, sales planning departments, financial departments and enterprise managers must all participate in the forecasting process.
The boss must realize that the integration of supply and demand is a way for the company to run. For the integration of supply and demand and demand forecasting, corporate culture is more important than any flow chart or technical means.