Worldwide, manufacturers are feverishly exploring ways to maintain sales. This is particularly true for fast-moving and general consumer goods. But new opportunities beckon in distribution management.
By Dawid Janse van Rensburg, MD: CargoSolutions, a division of Cargo Carriers

One definite way forward is to ensure high availability of the product to the market, and especially to ensure high availability on the shop shelf. A major and often incorrect concern about high availability is that distribution costs will spiral out of control with more frequent deliveries.

Various implementations of the Theory of Constraints (TOC) based ‘pull’ distribution concept have provided more perspective on the matter. Implementing a demand driven supply chain will almost always reduce inventory in the system, and the transportation and handling costs need not increase substantially. The TOC portfolio of supply chain solutions is the brainchild of Dr Eli Goldratt, famous for many books on management related subjects.
Fundamental is the understanding that an increase in sales is the biggest leverage that any supply chain has. One logical precondition is that the growth in sales must be much faster than the growth in operating expenses to achieve that growth. Therefore, if the increase in sales can only be achieved by substantial additional spend on transport, handling and an additional investment in working capital as a result of increased inventories, then the value proposition is not valid.

Why has the world been slow in implementing true demand driven distribution systems? Simply put, traditional systems do not enable the conversion, and remain largely ‘push’ based supply chains as most actions and decisions are triggered by a sales forecast, rather than actual consumption. Whilst the IT systems world continues to refine and optimise forecast algorithms, a highly accurate forecast remains as elusive as ever.

The advanced forecasting modules existing today try to model the demand and to create a good answer to the availability question: what to hold at where and when. However, the forecasting mechanism, no matter how good, cannot accurately predict actual demand.
Fact is that the narrower the aggregation, the worse the answer becomes – meaning that the question, ‘how much of the product will be sold in total?’ will yield a much better answer than the question, ‘how much of the product will we sell at this specific location?’ This phenomenon stems from the fact that fluctuations at specific locations average out when aggregated over many locations (assuming they are independent events). If we predict the sales at 100 different locations, we might believe and average location will range from a low of 10 to a high of 25 units a day (1000 to 2500). If we ask the same question on the overall quantity that we need to manufacture, we will get a much more accurate answer – probably ranging from 1650 to 1850.

The important part of the ‘pull’ model for managing a supply chain is to keep the stock, before it diverges, at the point where the stocks can be used to supply many different destinations, and using a pull mechanism from the destination to replenish. This method guarantees that we keep the lowest stock possible to support the demand of the various consumption points.

To have the product available at different locations it is recommended to aggregate the stocks at the source and build a plant or central warehouse (PWH/CWH). For manufacturers, this takes the form of a plant warehouse, and likewise, if the organisation is a distributor, the entity is a central warehouse. Most of the stock is kept in this warehouse. According to the principles of statistics, this aggregation guarantees a more stable system than keeping it at the different consumption points.

A number of implementations locally and internationally have exposed huge potential benefits, with the most significant being a substantial increase in sales, and a dramatic decrease in inventory levels across the supply chain

At the consumption point the amount of stock is very limited. Once a certain consumption point sells a unit, the consumed unit is replenished as soon as possible from the PWH/CWH. When the transportation time from the PWH/CWH to the consumption points is extensive, a regional warehouse (RWH) might be needed between the PWH/CWH and the consumption points. A regional warehouse will serve as a consumption point to the PWH/CWH and as a central warehouse to the consumption points it is serving.

At every downstream location where stock items are to be managed, an initial assessment of the correct buffer size is required. These decisions depend on the order fulfillment strategy that the company embraces, and the initial emphasis is to ensure that every location that needs to keep stock to ensure high availability, has a stock buffer size defined. Basically the stock buffers must ensure that the system can maintain supply within reasonable variation levels, and will be determined by demand at each location.
Obtaining real consumption data from stock locations such as retail outlets is also rapidly becoming a reality, and using basic building blocks of IT connectivity, data extraction routines and automated e-mail facilities, such data can serve as triggers for replenishment on an ‘as frequent as is required’ basis.

Using data triggers for replenishment, lead times can be shortened dramatically – order lead time is reduced to zero, production lead time is zero due to the aggregated finished goods stockholding policy, and essentially it is just transportation lead time that remains.
Next, a frequent mode of replenishment is vital. Dependent upon the nature of the product, frequent replenishment is essential to maintain a high level of availability at sensible inventory levels. For most FMCG distribution systems, this is already a feature of the system. However, very often certain policies and procedures limit the number of products to be replenished regularly, with pre-determined replenishment cycles for typically the longer shelf life products, or the slower movers.
Lastly, the buffer sizes must be dynamically managed to ensure that they reflect an optimum inventory position considering the replenishment regime, market demand and the variability in demand and replenishment. In TOC these dynamic buffer management algorithms have been refined over years of practical experience in a variety of different supply chains, and are perhaps the biggest feature of such pull distribution systems. The algorithms adjust buffer sizes on the basis of too much or too little signals, with the objective to maintain an optimum level of inventory of every item at every stock location.