Hello TOC

Finance & Measurements, as Taught by Eli Goldratt · Lesson 7 of 8

The five steps as the judge

If the strength of the chain is set by its weakest link, judge every decision by what it does to that link. Goldratt's five focusing steps, then two judgments rebuilt on them: which product to promote (throughput per constraint unit), and whether an investment pays (ΔT, ΔOE, ΔI), tested on five scenarios.

24 min · Not completed yet

Strengthen the weakest link

In the throughput world, we want to make the chain stronger, and the strength of a chain is set by its weakest link. So we need a mechanism that directs us to strengthen the weakest link. Goldratt’s mechanism has five steps:

  1. Identify the system’s constraint. It can be physical (a bottleneck), organisational (a department), or external (the market).
  2. Decide how to exploit the constraint: get the most out of what is already there. Providing more usually costs money; using what we have better usually doesn’t. A bottleneck should first be made 100% effective.
  3. Subordinate everything else to that decision. This means two things: manage all other resources so the constraint can be exploited, to protect throughput; and make sure they don’t do more than the constraint needs, to control cost.
  4. Elevate the constraint: add capacity, hire, buy. Only as the fourth step, because it usually costs a lot and exploiting is often enough.
  5. If a constraint has been broken, go back to step 1, and don’t let inertia become the constraint. When the constraint moves, every decision about exploiting and subordinating must be re-examined: the whole company must be re-oriented.

In operations, the first three steps give Drum-Buffer-Rope: identify the bottleneck; set the drum so it works on the right things in the right order; set the buffer and tie the rope so everyone else works on what it needs, and no more.

For the experienced: TOC recognises two kinds of constraint: things that limit the system’s throughput, and things that determine its behaviour, usually a wrong policy. The five steps handle the first kind. When one of the steps is hard to carry out, use the thinking processes to find the second kind.

The five steps aren’t only a process of ongoing improvement. They are the mechanism for judging how local decisions affect the whole company.

Judging product viability

A product doesn’t have its own profit; only an organisation does. So the question isn’t “what does this product earn?” but “what does promoting it do to the company’s profit?” The company’s throughput is set by its constraint, so: identify the constraint, then see how promoting the product affects its exploitation.

Throughput per constraint unit = (sales price − raw material) ÷ time the product needs from the constraint.

In the P&Q plant, B is the constraint. P earns $45 of throughput and needs 15 minutes of B; Q earns $60 and needs 30 minutes. Which product should be promoted?

  • Q: it earns more throughput per unit sold
  • Either: they’re about equal
  • P: $3 per minute of B, against $2 for Q

Every minute B spends on P brings $3; on Q, $2. That is why P first made $300 a week while Q first lost $300.

Notice what this calculation is not: it isn’t another way to calculate product cost or to allocate anything. It has nothing to do with cost. And it is much easier: product cost needs data from every work centre plus overhead factors; throughput per constraint unit needs prices, material costs and data from one resource. It is the right way, and the easy way.

Judging an investment

For any investment, ask what it changes in all three measures:

ΔROI = (ΔT − ΔOE) ÷ ΔI

Here are five investments in the P&Q plant. Work out the change in throughput for each before opening the answer. (Today: 100 P and 30 Q a week; B is the constraint.)

Scenario 1: an investment gives B 10% more capacity. What is ΔT per week?

  • $0
  • $480
  • $1,200
  • $600

B had 2,400 minutes; now it has 240 more. All the P demand is already met, so the extra minutes make more Q, which the market will buy: 240 × $2 = $480. The gain from extra constraint capacity must take the available market demand into account.

Scenario 2: an investment (say, a promotion) raises demand for P by 10 units a week. What is ΔT?

  • $450: ten more P at $45
  • $0
  • $150
  • $300

Ten more P use 150 minutes of B, at the expense of Q. Each of those minutes now earns $3 instead of $2: 150 × $1 = $150. An investment can raise throughput not only by adding constraint capacity, but by enabling better exploitation of the constraint.

Scenario 3: a second B machine is bought, doubling B’s capacity. What is ΔT?

  • Double: $6,300
  • $2,400
  • $480 × 10
  • $1,200

B is no longer a bottleneck; every resource can meet all demand, and the constraint moves to the market. Q goes from 30 to 50: 20 × $60 = $1,200. Extra capacity can break the constraint, and then the gain is limited by the new constraint.

Scenario 4: an investment lets P use an alternative purchased part costing $2.50 instead of $5. What is ΔT?

  • $250
  • $0: nothing more is sold
  • $500
  • $125

Throughput is sales minus raw material, so it also rises when material cost falls: $2.50 × 100 P = $250.

Scenario 5: an investment gives resource A 20% more capacity. What is ΔT?

  • 20% of throughput
  • $0
  • $480
  • $250

A isn’t the constraint; nothing more can be sold. ΔT = 0. The investment might still be justified, but only through the other two measures.

ΔOE. Two warnings. Reductions in OE are often justified by “saving man-hours”; if nobody will be laid off, it is just a numbers game. Goldratt’s test: ask for the names of the people who will leave. And new equipment often raises OE: maintenance, support, power.

ΔI. Count the price of the equipment, and also the change in inventory the company must hold. Adding capacity at the bottleneck reduces the protective capacity of the others, so the buffer (inventory) or their capacity must grow. Adding capacity at a resource that causes most of the holes in the buffer lets the buffer shrink, and inventory falls.

Test yourself: Helen approves a CAD-CAM system that makes the engineers more efficient. Which measures does it change, and when?

  • It always raises T, since engineers produce more drawings per week with the same effort
  • It only reduces OE
  • T rises only if engineering is the constraint or faster engineering wins orders

Check all three, every time.

Check your understanding

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

1. Why is “elevate” only the fourth step?

  • It is the least important step, kept for when all else fails
  • It usually costs a lot, and exploiting the constraint is often enough
  • It is illegal before step 3
  • It never works

Use what you have before buying more.

2. Subordinating non-constraints means…

  • Keeping them fully busy, so they never hold up the constraint
  • Cutting their budgets
  • Ignoring them
  • Making sure the constraint is exploited, and doing no more than it needs

Protect throughput and control cost.

3. Throughput per constraint unit needs data from…

  • Prices, material costs and the constraint only
  • Every work centre and overhead factors
  • The general ledger
  • Competitors

Right and easier.

4. Doubling B broke the constraint. The gain was limited by…

  • B’s new capacity, which was still not enough
  • Resource A
  • The market demand for Q, the new constraint
  • Operating expense

Then go back to step 1.

5. Goldratt’s test for an OE saving based on man-hours is…

  • Ask the cost accountant to check the calculation
  • Ask for the names of the people who will be laid off
  • Add 10%
  • Accept it

Otherwise it’s a numbers game.