The Goal

The problem it solves

Most teams measure how busy each person or machine is, and most managers push those numbers up. The Goal argues that this is the wrong target. In its plant, everyone is working and orders are still weeks late, stock is piling up and the bottom line is poor. If your own business has a backlog next to a team that never has a spare hour, this is your situation described in a factory.

What changes after reading it

You stop asking whether a step is efficient and start asking whether it limits the whole. You look for the one place where work waits, and you treat time there as the most expensive time you have. You also get a reason to leave some resources idle on purpose, which is hard to accept until the plant in the story shows what happens when you do not.

Why a novel works here

Goldratt says he chose the form so that readers would deduce the answers themselves, with the mentor Jonah asking questions instead of handing out conclusions. That is why it sticks. You carry the story of the slow scout on the hike and the dice game into your own meetings, and you can explain them to a colleague in two minutes.

How it applies outside a factory

The author says he hoped readers would see the same principles in banks, hospitals and insurance companies. The interviews added to this edition back that up with managers describing results in a bank, a hospital and schools, as well as in manufacturing. Goldratt himself says service organisations differ so much from each other that covering them would take a library, so use the examples as a way of thinking. They are not a recipe.

When to read it, and when not

Read it before you hire, buy software or add a shift to fix lateness. Read it again when a fix works and a new problem shows up somewhere else. Skip it if you want a short manual: it is a long novel with a family subplot, and the method has to be pulled out of the story. The numbers and tools of the 1980s plant are dated, the reasoning is not.

Decisions and repeatable work

The loop in the book is a good habit for logging decisions: write down what you think the constraint is, what you decided to do about it, and what changed. Once you know where the limit sits, the work around it can be turned into a written routine, so that the limiting step is always fed first and never waits on someone's memory.

The argument

The Goal argues that a business exists to make money, that most of the numbers managers chase work against this, and that the output of any system is set by its constraint. Goldratt teaches it as a novel, and says in his introduction that the mentor supplies question marks instead of exclamation marks so that readers reach the answers first.

A plant with three months left

Alex Rogo runs the Bearington plant for UniCo's UniWare division. His boss, Bill Peach, turns up unannounced, demands the status of a customer order that is seven weeks late and gives the plant three months to turn around or be recommended for closure. The same morning a master machinist quits after Peach threatens to fire him, and the NCX-10, the only machine of its type in the plant, is left damaged. Everything is late, inventory is high and the efficiency reports look fine. Layoffs and a new line of robots have not saved the plant.

What the goal really is

In an airport Alex meets Jonah, a physicist he used to know, who asks whether the robots raised sales, cut inventory or cut expense. They had done none of the three, which Alex and his staff later confirm from the records. Jonah leaves him with one question: what is the goal? Alone on a hill above the plant, Alex tries technology, quality and efficiency as answers and settles on one: to make money. With his controller, Lou, he settles on net profit, return on investment and cash flow as the measures of that goal, and sees that they say little about a plant's daily decisions.

Three measures instead of cost accounting

Alex phones Jonah, who gives him three measures. Throughput is the rate at which the system generates money through sales, inventory is the money invested in things it intends to sell, and operational expense is the money spent turning inventory into throughput. Production that is not sold is not throughput. Alex and his staff test the definitions on awkward cases such as knowledge or the chairman's chauffeur. The robots fail the test: to keep them busy the plant released more material, so inventory and carrying costs rose while sales stayed flat, and the better efficiency figure was working against the goal.

Herbie and the dice game

At breakfast in New York Jonah says a plant where everyone is working all the time is very inefficient, and that trimming capacity to match demand makes throughput fall and inventory rise. The cause is two things found in every plant: dependent events, where one step must finish before the next can start, and statistical fluctuations, where each step varies. Alex works it out on a scout hike with his son. The line of boys stretches out because no one can go faster than the boy ahead, and the slowest, Herbie, sets the pace of the whole column. Alex moves Herbie to the front, takes his pack off him, and the column stays together. At a picnic stop he models a perfectly balanced plant with matches, bowls and a single die. After ten rounds about twenty matches had come through instead of the thirty-five expected, with stock piling up at the end of the line. Slowness accumulates down a chain of dependent steps instead of averaging out.

Bottlenecks and the rule that an hour is not an hour

Jonah's first rule is "Balance flow, not capacity." He defines a bottleneck as any resource whose capacity is equal to or less than the demand placed on it and tells the staff to find their own. They do, after a slow search through the data: the NCX-10 and the heat-treat furnaces, which Bob Donovan calls the plant's Herbies. When Jonah visits he asks how these two are run and finds lunch breaks while the machine stands idle, defective parts rejected after the bottleneck instead of before it, and piles of parts that will not ship soon. He then shows that an hour of bottleneck time does not cost its standard rate of $32.50. It costs the plant's whole monthly expense divided by the bottleneck's hours, about two thousand seven hundred dollars. In his words, "an hour lost at a bottleneck is an hour lost for the entire system." He also teaches that "activating a resource and utilizing a resource are not synonymous", so running a non-bottleneck flat out only builds stock, and optimising every part separately does not optimise the whole. In a later call he adds that an hour saved at a non-bottleneck is a mirage.

The first changes on the floor

The staff put the idea to work. Quality inspection moves in front of the bottlenecks, so they only work on good parts. Crews are stationed at the NCX-10 and the furnaces on every shift, and the break rules change. The Zmegma, an old machine, and two other old ones are brought back for one shift a day to take load off the NCX-10. Some heat-treat goes to an outside vendor, and some of it turns out to have been specified by the plant itself and can be dropped. Parts that need a bottleneck get red tags and priority, the rest get green, and the most overdue orders are worked first. Within weeks the plant sets a record for shipments and the backlog of late orders begins to fall.

The constraint moves

New shortages appear, this time of green-tagged non-bottleneck parts that never reach final assembly. Stacey, who runs inventory control, suspects new bottlenecks. Jonah looks at the floor and says the milling machines are not bottlenecks. The plant had been releasing material whenever non-bottlenecks ran short of work, so the bottleneck work loaded those machines and the green parts waited for weeks. The fix is to release material for the red parts only at the rate the bottlenecks need it. Alex's children suggest a drummer and a rope for the scout line, and the plant does the same with a schedule built by the data processing manager, Ralph Nakamura: material goes in about two weeks before it is due at a bottleneck, with a few days of stock waiting there as protection. Efficiencies dip while the excess stock is used up, then recover. Inventory falls and ship dates become predictable to within a day.

Smaller batches and shorter lead times

Jonah explains that a part's time in the plant splits into setup, process, queue and wait, and that queue and wait take most of it. The bottlenecks decide both. That puts the standard economic batch formula in doubt, and he asks Alex to halve the batch sizes. Bob objects that this doubles the setups. The answer is the same rule: time saved at a non-bottleneck is a mirage, because those machines now have idle time anyway. Lead times had already fallen from three or four months to about two. With halved batches the staff expect about four weeks, and sales begins to quote six. On paper the cost of parts rises because of the extra setups, although actual expense has not moved.

The Burnside order and the results

Bucky Burnside, whose late order opened the story, asks for a thousand Model 12s in two weeks. Alex halves the batches again, reschedules orders that were due to ship early and offers 250 a week for four weeks. Every delivery is on time, and Burnside lands by helicopter to shake hands with the workforce. At the end of the quarter the plant has beaten its fifteen percent profit target with seventeen percent, inventory is down to about forty percent of what it was and throughput has doubled.

Accountants and a promotion

Then a camera crew sent by Hilton Smyth, the division's productivity manager, sees the small batches. An audit team led by Neil Cravitz rejects the way Lou changed the base for product costs and recalculates profit at 12.8 percent. At the review Alex argues that the bottom line is the test, and he takes his case to Peach. The plant is not closed. Peach, the divisional controller Ethan Frost and the marketing head Johnny Jons are moving up to the group, and Alex is told he will run the division. Lou later finds that valuing inventory at cost had hidden real profit of over twenty percent. Along the way Julie leaves for a time, and she and Alex slowly rebuild their marriage.

The five steps on the board

Alex and his team write down what they have been doing as a process: identify the constraint, decide how to exploit it, subordinate everything else to that decision, raise the constraint's capacity, and go back to the start once it is broken. The first draft says bottleneck, and the team swaps it for constraint after Stacey points out that the constraint has already changed from machines to the way material is released and then to the market. Stacey's correction to step five adds a warning not to let inertia create a new constraint. Inertia had, in fact, kept the plant loading its bottlenecks for orders that did not exist.

The market becomes the constraint

With production flowing, the plant has about twenty percent spare capacity, so Alex needs orders. Jons says orders are scarce and offers none, apart from a French buyer asking for nearly the whole range at a price below the plant's full product cost, though about twice the material cost of a Model 12. Alex takes it: with spare capacity the only extra cost is materials, and the deal adds a seven-figure contribution to net profit. Two weeks later a dozen work centres are on overtime and orders slip. The new orders had not overloaded any machine. They had eaten the spare capacity the non-bottlenecks need to rebuild the stock in front of the bottlenecks after a disruption. Bob, now running production, extends the release time to two weeks and sales is told to promise nothing under four weeks. A constraint can reappear when the protection around it is too thin.

From a plant to thinking processes

As Alex faces a whole division, he sees that the real constraints are often policies and measurements, and even the machine bottlenecks had been policies in disguise. He reasons that a manager must answer three questions: what to change, what to change to, and how to cause the change. Reading popular science, he notices that physicists reason with cause and effect, if this then that, and Julie says the Socratic dialogues work the same way. Jonah will not keep advising him and asks him first to say what he wants to learn. His finance colleague Lou says they should learn to be their own Jonahs.

An Interview with Eli Goldratt and Others

The third revised edition ends with an interview by David Whitford of Fortune Small Business. Goldratt answers whether the book is still relevant with a test: implement the knowledge in one plant of a large organisation and compare its results with the rest. Interviewees from General Motors, the Marine Corps, a book manufacturer, an office supply business, a bank, a hospital in Pretoria and an education charity describe how they applied the book's ideas. Goldratt says the principles carry over to service organisations, though the detailed procedures still have to be developed for each.

What to do with it

• Write down the goal of your business in terms of money, then list the measures your teams are rewarded for and check each against it.
• Find where work waits longest and most often, and treat that step as your constraint until you can show it is not.
• Protect that step: feed it first, never let it starve, check its work before it reaches it and put your best people on it.
• Stop releasing work faster than the constraint can handle, and cut batch sizes to shorten the wait.
• After each improvement, look again for the new constraint, including a policy or the market, before you assume the job is done.