Wednesday, June 04, 2008

Achieving Distribution Excellence

Achieving Distribution Excellence

What Is Distribution?
Distribution is the function that moves items from their source through layers of diverging nodes to the point where they are purchased by the end users. An organisation may have an integrated supply chain, where the manufacturing and distribution functions are within the same business. Alternatively, the distribution function may be one or more separate businesses.

The mandate of distribution is to ensure that end users have what they one in the quantities they want, where they want it and when they want it. Unfortunately the clarity of the requirement is not matched by the effectiveness with which it is fulfilled.

Typical Distribution Blues
You go into a store to ask for a particular design of tiles. It is not available in stock. The store has lost a sale and you go away disappointed. Meanwhile that same store has several pieces of another item that they are unable to sell and might have to dump.

Talking of dumping, this is exactly what happens when fashion shops declare clearance sales when the new season's designs are being expected. When items are sold at deep discounts of up to 50% of their original price, you can be sure no profits are being made on them. These same stores will often proudly hang the label "Sold Out!" with respect to other items, without realising the implication that if they had had these items in stock, the store would have made a higher profit.

In geographically dispersed distribution chains, fast moving items in on region will usually be slow moving in others, so that there are shortages and surpluses of the same goods in different areas. An attempt is made to bridge the gap by transferring between regions. Transportation costs go up.

The Cadillac Miracle
Before 1999, the Cadillac distribution network suffered from all the "blues" outlined above. The case was particularly complex as they has on offer about 7.5 million legitimate vehicle configurations they could offer customers, each containing about eight thousand parts bought from about one thousand suppliers.

Of course no single dealer could hold anywhere near the total possible car configurations: the largest dealers might have about 200 cars on their lot, made up of say forty different configurations. The dealer could (usually reluctantly) get extra configurations from other dealers within a 500 mile radius - at extra transport cost and reduced profits. Even so, typically only about 1000 configurations would be available within the 500 mile radius.

If a buyer insisted on a configuration not held within the area, it had to be ordered straight from the plant. This might take three to four months to deliver. In forty percent of the cases, the car delivered did not conform to the customers specs!

In 1999 Cadillac implemented a radically different distribution system. The result is that in the last eight years delivery time for specially ordered configurations has shrunk from over three months to nineteen days or less! And their inventory within the whole supply chain reduced by as much as fifty percent. How did they do it?

The Distribution Conflict
The distribution function - whether it is a separate business or part of an integrated supply chain - is pulled in two opposite directions in its attempt to manage well.

In the first case, there is the desire to ensure no potential sales are lost. This means that stocks must cover maximum "foreseen" demand within the time it takes to replenish the stocks, plus an allowance for fluctuations in supplier reliability. Put briefly, in order to protect through put by selling all we can, we must hold large stocks.

On the other hand, we must control costs. Controlling costs (of storage, spoilage, and obsolescence) implies holding low stocks.

Just how do we go about improving availability to the customer while reducing inventory requirement? To repeat the question we asked before, how did Cadillac do it?

Rethinking Distribution
Obviously, the solution must involve a radical rethinking of distribution as a whole. To paraphrase Einstein - the significant problems we face cannot be solved at the same level thinking that led to the problems in the first place.

Conventional wisdom and practice says that stocks should be held as close to the end user as possible. Most supply chains attempt to do just this. The result is the paradox of simultaneous product unavailability combined with excess inventory. Why?

Recall the rule (stated above) for deciding how much stock to hold. Part of it has to do with "foreseen" demand. How accurately can a store forecast sales of a particular item for a given period? So poorly as to make such forecasts meaningless. While forecasts for individual stores are highly unreliable, aggregate forecasts for several stores together have a far higher level of reliability. Mathematically, accuracy of forecast improves by the square root of the number of individual nodes aggregated. For example sales or demand forecasts for a wholesaler or regional warehouse supplying one hundred retail points is ten times as accurate as the forecast for any individual retail point.

What did this mean for Cadillac and what does it mean for your supply chain? It means rather than concentrate stocks at the points closest to consumption, they are better held at points of highest forecast accuracy. Then any items sold from any retail point are immediately replenished.

The same arrangement exist between the all the other layers in the distribution chain, back to the production plant. At the production plant, a plant warehouse forms the first layer in the chain. Production from the plant just replenishes items withdrawn from the plant warehouse and no more.

The radical improvements in Cadillac's distribution was achieved by applying approaches invented by Goldratt - the so called Theory of Constraints.

No comments: