Distribution & Supply Chain, as Taught by Eli Goldratt · Lesson 4 of 7

The plant warehouse

Is a forecast equally accurate everywhere in the chain? No: it is far more accurate where demand from many points adds up. So, against every instinct, hold most of the stock at the source. Karan's sceptical conversation with Ruth, step by step, and a quick check of whether a company has really implemented it.

24 min · Not completed yet

Is the forecast equally accurate everywhere?

Goldratt asked a question nobody asks about forecasts: is the relative accuracy of the forecast the same at a supply point as at a point of consumption?

The demand on a supply point is the sum of the consumption of all the points it feeds. Random ups and downs at different shops cancel each other out. So:

The relative variability of demand at a supply point is much smaller than at a point of consumption.

Distribution is built on the wisdom of holding inventory close to the customer. But now we see that the most reliable place to forecast is the supply point: the further from the end consumer, the more reliable the forecast. What if, against all instinct, we held most of the inventory not at the points of consumption, but at the source of supply: at the plant?

Karan’s objection

Karan: “If we want goods immediately available to customers, we should hold them away from customers? You’ll have to explain that.”

Ruth: “Each point must hold its target: the maximum consumption within its replenishment time, factored by unreliability. Followed everywhere, that leads to too much inventory. My claim is that adding a plant warehouse reduces the total inventory enough to solve the problem.”

Karan: “I still don’t get it. You just added inventory at the plant. Where’s the saving?”

Ruth: “In every regional warehouse. Before, their replenishment time was production plus transport. Now the stock they need sits in the plant warehouse, so their replenishment time is only the transport time.”

Karan: “So you’ve split the inventory between the regions and the plant. Big deal. Where are the savings?”

Ruth: “Remember the ‘maximum’. The lower the variability, the smaller the maximum you need. At the plant warehouse, demand is the sum of all the regions, so its variability is far lower. The maximum the plant warehouse needs is much smaller than the sum of the maximums the regions needed. What we hold there is less than what we removed from the regions.”

Karan: “That’s how you use the better forecast at the supply point.”

Ruth: “That’s one part. Forecasts also deteriorate with the length of time forecast. The regions now forecast only over the transport time, so their maximums shrink too. And there’s much less chance of a shortage in one region while another sits on surplus of the same product.”

The second benefit: production lead time

Ruth: “Most of the production lead time is the time an order waits in the queue. What fills the queue with urgent orders today?”

Karan: “Any shortage in any regional warehouse triggers an urgent production order.”

Ruth: “How often is an urgent order triggered because one warehouse is out of an item that the system has plenty of somewhere else?”

Karan: “For at least half the products, I’d say.”

Ruth: “With a plant warehouse, an urgent order is triggered only when the plant warehouse itself is out. That means far fewer urgent orders and a drastically shorter production lead time, which saves more inventory at the plant warehouse. And without all those emergencies, running the plant becomes much easier.”

The third benefit: reliability

Ruth: “The formula has one more part: unreliability of replenishment. For the regions, it is now only the unreliability of transport, usually much smaller than that of the supplier, so they need even less. And for the plant warehouse: what makes a supplier unreliable?”

Karan: “Mostly, other orders jumping the queue.”

Ruth: “And we just removed most of the urgent orders that jump it.”

Karan: “So you’ve met all three conditions: a more accurate forecast, shorter replenishment, more reliable re-supply.”

Ruth: “Without a new forecasting system, without paying for speed, and without replacing a single vendor.”

Summing up: what does introducing the plant warehouse do?

  • It adds inventory and adds a step, so it only makes sense when transport is very cheap
  • It reduces total inventory while improving service in the regions
  • Only reduces transport cost

A significant portion of the stock is kept where demand is least variable, and less inventory is pushed into the system on a dubious forecast.

When the plant won’t join: a central warehouse

If the plant isn’t part of the distribution system and can’t be convinced to join, most of the benefits can still come from a central warehouse. Like the plant warehouse, it decouples the supply point from the points of consumption. The only benefit that may be lost is the effect on production lead time.

Have you really implemented it?

Having a central warehouse doesn’t mean the solution is implemented. Most distribution systems still push inventory down to the regions, even when they have one. Implementing the solution means more inventory in the central warehouse than in all the regional warehouses combined. Why?

Consumption from the central warehouse equals the consumption of all the regions together, so that isn’t the difference. The central warehouse has lower variability of consumption but higher variability of replenishment, and those roughly compensate. The big difference is replenishment time: replenishment to the central warehouse is usually many times longer than replenishment to the regions. So the ratio of central inventory to all regional inventory should follow the ratio of their replenishment times.

Case 1: replenishment to the plant warehouse takes one month; transport to the regions takes about 3 days. Roughly, how should inventory be split?

  • Equally
  • Three times more in the regions, close to where it is sold
  • About ten times more in the plant warehouse than in all the regions

30 days against 3. If a company in this situation has the opposite ratio, it hasn’t implemented the solution.

Case 2: goods come from the Far East to a central warehouse in Europe in 6 weeks; regions are about 1 week away. The ratio?

  • About 6 to 1 in favour of the central warehouse
  • About 1 to 6, since the regions are closer to the customers
  • Equal

Inventory at each location in proportion to the replenishment time to it.

Check your understanding

Answer all five. Get four right to complete the lesson.

1. Relative variability of demand is lowest…

  • At a single shop
  • At a regional warehouse, which serves a defined territory
  • At the supply point that feeds many points
  • Everywhere the same

Random fluctuations average out.

2. With a plant warehouse, the regions’ replenishment time becomes…

  • Only the transport time
  • Production plus transport
  • Longer
  • Zero

The stock they need is already made.

3. Production lead time falls because…

  • The plant works faster once it no longer makes stock for forecasts
  • Batches grow
  • Forecasts improve
  • Urgent orders are triggered only when the plant warehouse runs out

Fewer orders jump the queue.

4. Implementing the solution means the central warehouse holds…

  • Less than the regions
  • More than all the regions combined, in line with replenishment times
  • Nothing
  • The same as each region, so stock is spread evenly

Most companies push it downstream instead.

5. If the plant won’t take part, most benefits can come from…

  • More regional stock
  • Better forecasts
  • A central warehouse
  • Bigger trucks

Only the production-lead-time benefit may be lost.