Forecasting

Managing Lead-Times and Demand

It is the projected level of demand for a product from consumers during its lead-time, from the supplier to the retailer. Failure to estimate the level of lead-time demand for a given product can result in inventory shortages that cause customers to become dissatisfied while waiting for back ordered products. They may buy the product from a competitor instead, leading to lost revenue. Safety stock, if stored in sufficient quantities, may be used as a buffer in the event that lead-time demand is misjudged. Overestimating lead-time demand can lead to inefficient use of storage space and is not consistent with lean supply chain management practices.

Forecasting Future Demands

A forecast is an estimation of future demand. Most forecasts use historical demand to calculate future demand. Adjustments for seasonality and trend are often necessary. Forecasting is all about turning unknowns into knowns (or reasonable approximations).

Whilst recognising that each item within the inventory system is identified by its description and/or item code has its own individual set of characteristics; it would not be possible to devise a different set of processes and controls for each item.

Conversely, in setting up an inventory system for a range of widely-different items, it is obvious that no system of control can be applied to the entire range.

It is necessary, therefore, to have a process for classifying items into groups/types to enable the relevant system to satisfy the investment cost and service level goals to be identified and applied.

In reviewing the item range in any inventory system, some/all of the following elements need to be considered to predetermine the system to be applied.

Random/Predictive Demand

This initial item classification is usually carried out by firstly identifying any item which has a predictive demand.

By process of elimination, the rest of the item range will be classified as having a random demand.

It is necessary to set precise rules for the identification of predictive demand items. In this context, predictive relates only to those items that have a quantity and time commitment for when they will be required. These would include:

• Items that are called off on a scheduled basis by a customer, with no deviation in quantity or time, against the original forecast.
• Items for a sales campaign or promotion, will cease on the sale of the initial supply and not generate any further demand.
• Items are provided in advance of a new product launch, sales campaign, or promotion; note that these items may later become random demand items.

These do not include any items with any uncertainty in the demand – this is the criterion.

Any item where there is uncertainty in the demand will be classified as random.

Stable Demand

A stable demand pattern is one where, although the demand rate varies, it varies at a constant average over time. As such, it will provide no evidence of an increasing or decreasing trend.

Trend Demand

A trend demand pattern is one where the average demand rate varies over time, showing a tendency to increase or decrease.

In reviewing the general pattern of the demand, care needs to be taken in establishing the time period over which the assessment is made to ensure that the trend slope is sufficiently established.

Seasonal Demand

A seasonal demand pattern is one which shows a variation in the average demand at different points in time throughout the planning cycle, and can generally be related to market forces, which influence the demand patterns.

The periods of seasonality can vary from very short periods of one or two weeks to periods covering longer periods, e.g. three or four months.

In addition, seasonal demands can be expressed in two forms. The first shows an evolving pattern of development into and out of the seasons. The second is typified by a stepped function change in demand patterns for each of the seasons.

Where the inventory system contains a wide range of items, a full analysis, on an item-by-item basis, can be daunting. Inventory management will need to apply its knowledge and experience in establishing a category for each item, and use sampling techniques for groups of similar items, to reduce the level of activity to manageable proportions. Where the pattern is not obvious from the available data, the use of graphical representation can often provide an answer.