Measuring and synchronising a supply chain

Measuring execution

Throughput, investment and operating expense judge the plan: the policies and decisions. But even good decisions give poor results if execution is poor. How should execution be judged?

Goldratt’s answer: to drive fast improvement, focus on what is not done properly. Praising what is done well is good for morale, but it doesn’t trigger corrective action. Highlighting what wasn’t done well does, and showing improvement against the previous period provides the motivation.

Two kinds of “not done properly”:

Reliability: things that should have been done and weren’t. The end result is a missed commitment to a client. Missing something the client would pay $10,000 for isn’t the same as missing something worth $100, and a day late isn’t a month late. So:

Throughput-dollar-days = Σ (value of each commitment not met on time) × (days late).

Value means the selling price to the next link (or, better, the price of the final product to the end consumer). Days are counted from the due date until delivery, or until today. The target is zero.

Effectiveness: things that shouldn’t have been done but were. The end result is excess inventory, and what matters is its value and how long it sits. So:

Inventory-dollar-days = Σ (value of inventory on hand) × (days since it entered this link’s responsibility).

For simplicity, measure all inventory, not just the excess, and don’t aim for zero: aim for the minimum that ensures reliability.

Efficiency. Add local operating expense: only the expenses this link controls, such as its salaries and scrap, with no allocated corporate cost.

Every link: primary measurement throughput-dollar-days, target zero; secondary inventory-dollar-days and local operating expense, to be reduced, but never at the expense of the primary.

The power of throughput-dollar-days: suppliers

Even when customer and supplier computers are linked, there’s always an informal system alongside: the customer’s buyers on the phone, chasing parts. Why?

Now suppose the customer judges suppliers on throughput-dollar-days and sends them a daily status report, valuing each late item at the value of the end product it holds up. Suppliers will expedite the right orders on their own, without waiting for frantic calls that usually come too late.

What it does to a plant that feeds distribution

Today such a plant’s master schedule follows forecast-based orders, changes constantly with urgent orders for products short in even one warehouse, batches heavily, and expedites by the level of screaming. It is hectic.

With the distribution solution, the master schedule is built to fill the plant warehouse to its targets. Throughput-dollar-days guide the priorities; inventory-dollar-days stop excessive batches; and work is scheduled and prioritised by Drum-Buffer-Rope and buffer management.

The measurements also change how the plant treats the improvement programmes corporate keeps sending it:

• Lead time. Under cost measurements, a shorter production lead time does almost nothing for the plant’s numbers, so it feels like another programme imposed from above. Under dollar-days, shorter lead time cuts throughput-dollar-days and inventory-dollar-days. The plant starts reducing lead time without being asked.
• Set-ups. Under cost measurements, plants save set-up cost by enlarging batches, and set-up reduction goes where it’s easy, not where it matters. Under dollar-days, faster set-ups on resources without enough protective capacity improve both measurements, so the effort becomes focused.
• Quality. Today quality hurts the plant’s numbers only in extreme cases. Under dollar-days, when a downstream link finds a quality problem, the commitment reopens until it is fixed, and the later the problem is found, the bigger the penalty. “Quality at the source” stops being a slogan.

Dollar-days measurements induce and focus every local improvement effort. Goldratt suggested they should be the operational measurements of any company implementing Lean.

Implementing the measurements. Goldratt’s warning: “Change the measurements to new ones I don’t fully understand, and nobody, including me, knows how I’ll behave.” So guide people to discuss the cause and effect between the new measurements and: conventional efficiencies, large batches, long lead times, long set-ups, and quality that makes clients complain.

Synchronising a supply chain

Synchronising a supply chain is an order of magnitude harder than synchronising a company. Inside a company, when local optima (a department’s performance) conflict with the global optimum, the local optima should be discarded. In a supply chain of separate businesses, neither can be discarded: competition increasingly happens between supply chains, so the chain’s performance is vital to every member; but each company’s goal forbids it to sacrifice its own performance for the chain.

In a company, a conflict between local and global shows the measurements are wrong. In a supply chain, it shows the rules governing the business relationships between the companies are wrong. New rules must answer two questions for each link:

When does a sale actually take place? Today, when goods are shipped or accepted. But every top manager knows that until the end consumer buys, no one in the chain has really sold. So internal transfers of inventory should not count as sales. A sale happens when the end consumer buys, and that money is shared among the links that supplied the goods. (A prerequisite: shrink supply lead time below today’s payment terms, or find a way to finance upstream inventory.)

What is the value of the sale? Price is set more by the market than by the supplier, and the link closest to the customer knows best what different customers need. So businesses should negotiate not a fixed price, but a percentage of the price(s) to the end customer, whatever they turn out to be, with a minimum.

The internal supply chain

A chain of entities inside one company should be easier, since local optima can be discarded. Yet many companies make it harder than a real supply chain, by treating their subsystems as profit centres. That has two ramifications. The subsystem is judged as if it were a stand-alone business, so local optima can no longer be discarded. And transfer prices must be set, usually from cost, which makes local optima inside each profit centre matter too. No wonder so many profit centres fight to buy from outside rather than from sister units.

Don’t treat a subsystem as a stand-alone business unless it is allowed to be one, free to choose whom it buys from and sells to. Stop measuring subsystems as profit centres.

Checklist and results

1. Establish the plant (central) warehouse.
2. For each product at each location, set the inventory target by the formula.
3. Move to “order daily, replenish frequently”.
4. Monitor the targets by the colour zones.
5. Re-examine make-to-stock and make-to-order policies.
6. Teach each link to monitor execution with dollar-days measurements.

Implement on a significant product line first; clear results should appear within three months. Goldratt’s typical results:

• Inventory in the system down, typically by 50%.
• Sales up, typically by 20%.
• Inventory turns more than doubled.
• Transfers between regional warehouses down to almost zero.
• Obsolescence down to less than half.
• Operating expense about the same.
• Relationships with suppliers and clients significantly better.

If after six months these haven’t appeared, he advised calling in an expert: it may be your unique environment, but more likely a significant implementation mistake.

Check your understanding

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

1. Throughput-dollar-days measure…

• Inventory value over time at each link of the chain
• The value of commitments not met on time, times the days late
• Local expenses
• Sales per day

Reliability; the target is zero.

2. The primary measurement of every link is…

• Local operating expense
• Inventory-dollar-days
• Efficiency
• Throughput-dollar-days

The others are reduced only without hurting it.

3. In a supply chain, a sale really happens when…

• The end consumer buys
• Goods leave the plant
• The invoice is sent
• The distributor accepts the goods

Until then, nobody in the chain has really sold.

4. Instead of a fixed price, links should negotiate…

• Cost plus margin
• Volume discounts
• A percentage of the end customer’s price, with a minimum
• Annual contracts only, with prices fixed for the year

The last link knows the market best.

5. Goldratt’s typical inventory reduction from the distribution solution was about…

• 10%
• 50%
• 90%
• None

With sales up about 20% and turns more than doubled.