The Goal: A Process of Ongoing Improvement

Miles: A plant manager walks onto his own factory floor and finds a robot. It's new, it cost a great deal of money, and the department it sits in is posting the best efficiency it has ever posted. Every number on the report is up.

Miles: The plant is late on every order it has, the warehouse is full, and the division has given it three months before it closes.

Miles: Nothing is broken. Everything is working. That's the problem.

Miles: Today: The Goal, by Eliyahu Goldratt and Jeff Cox. In two sentences: what a system produces is set by its single slowest link, so improving anything else produces paperwork rather than goods. Thirty books in, we've been asking what a leader ought to be. This one asks what the work is doing while you're being it.

Miles: Sixty Books. Presented by Stoar Consulting and Coaching.

Miles: Book thirty one. You're in the reading room, this is Sixty Books, and I'm Miles. The arrangement holds: one management title an episode, sixty of the strongest of the last hundred years, each one asked the same two questions. What does it give a person who leads, or gets led, with a brain wired to a different plan. And how much survives a job that now has AI sitting inside it. Nora brings the evidence, and this time she built the episode out of it.

Miles: Nora, one thing before anything else. The most influential book ever written about a factory is a novel, and it has a subplot where the hero's wife walks out on him.

Nora: It does. And the way he gets her back is by running the same set of questions at the marriage that he'd been running at the plant.

Miles: Naturally he does.

Nora: It works better than it has any business working, and the reason isn't romance. The method is a way of asking what's actually limiting a thing, and a lot of things have a limit nobody has ever named.

Miles: Set the scene properly.

Nora: Alex Rogo runs a plant for a division of a company called UniCo, in a town called Bearington. The novel opens with his division vice president parked in Alex's space, inside the plant, holding a very late order and an ultimatum. Three months to turn the place around, or it shuts.

Miles: And the plant looks fine on paper.

Nora: Better than fine. Efficiencies are up, the automation is the pride of the division, and the plant can't ship. That gap is the entire book.

Miles: Enter the physicist.

Nora: Jonah, whom Alex knew at university, run into by accident at an airport. Jonah never answers anything. He asks two more questions and sends Alex back to find out for himself, and that's the structure of the whole novel. You don't get told the answer, you watch somebody derive it.

Miles: Start with the first thing Jonah makes him say.

Nora: What the goal of the company actually is. Alex tries quality, technology, market share, employing people. Jonah won't take any of them, and where they land is that the goal is to make money and everything else is a means to it. Which reads bleak and is doing real work, because if you can't state the goal you can't say whether any given action moved you toward it. This plant is full of actions everyone calls good and nobody can connect to anything.

Miles: So he replaces the measurements.

Nora: With three. Throughput, the rate at which the system makes money through sales, and the word sales carries that sentence. Inventory, the money tied up in things it intends to sell. Operational expense, what's spent turning the second into the first. Make something nobody buys and you've moved money between pockets and written it up as output.

Miles: Right.

Miles: And then the hike.

Nora: The one everybody remembers. Alex takes a troop of boys walking, the line keeps stretching out, and he works out why. Each boy's pace varies, and no boy can walk faster than the boy ahead of him. Fluctuation plus dependency. The quick ones can't recover the lost ground, because the gains don't accumulate and the losses do.

Miles: And Herbie.

Nora: The slowest boy. Alex puts him at the front, takes the weight out of his pack and spreads it round the rest, and the whole troop speeds up. The troop's pace was never the average. It was Herbie's.

Miles: And he goes back and finds Herbie in the plant.

Nora: Two of them. A heat treat operation and a machine they call the N C X ten. Everything else on that floor has more capacity than those two, and the plant's output has been equal to theirs the entire time.

Miles: The author, and he matters here more than most.

Nora: Eliyahu Goldratt, born nineteen forty seven in what was then British Mandatory Palestine. A physicist. Bachelor's from Tel Aviv University, master's and doctorate from Bar Ilan. His publisher's biography lists patents from medical devices to drip irrigation to temperature sensors, so he wasn't purely a theorist.

Miles: How does a physicist arrive at a factory?

Nora: Through scheduling. He worked at a firm called Creative Output, which built and sold production scheduling software described as the first to do finite capacity scheduling on a production floor, and The Goal came out of those years. The novel names no software and sells none, and that's as far as we'll take it.

Miles: And the book's own history.

Nora: Written with Jeff Cox, a novelist, which is why it reads at all. Published nineteen eighty four by North River Press, after people inside his own company and a run of publishers had passed on it. And one bibliographic detail almost everyone gets wrong. The subtitle you know, A Process of Ongoing Improvement, isn't the original. The eighty four edition went out subtitled Excellence in Manufacturing. The familiar wording arrives with the eighty six revision.

Miles: And it sold.

Nora: The publisher reports over five million copies and thirty five languages, which is their figure rather than an audited one. In twenty eleven, Time put it on a list of the twenty five most influential business management books. Goldratt died that June, at home in Israel, at sixty four.

Miles: Now the machinery.

Nora: Two claims, then the procedure. The first claim is what the whole book stands on. An hour lost at a bottleneck is an hour lost for the entire system. An hour saved at a non bottleneck is a mirage.

Miles: Say why the second half is true, because people take the first half and skip past the second.

Nora: Because a non bottleneck has spare capacity by definition. Speed it up and it makes more of something that then waits in front of the bottleneck. You've turned cash into a bigger pile of unfinished work and a better efficiency report, and output hasn't moved an inch.

Miles: So the correct move at most of those stations is to stop.

Nora: Sit down, yes, whenever there's nothing the constraint needs. Which is the hardest sentence in this book to say out loud on a factory floor, and it's why the argument had to arrive wrapped in a novel about a man's marriage rather than in a memo.

Miles: Huh.

Miles: And the second claim.

Nora: Jonah's line about the balanced plant. The closer you get to matching capacity exactly to demand at every station, the closer you are to going broke. Which sounds like a paradox and is arithmetic.

Miles: Go on, unpack it.

Nora: Here's where the novel meets mathematics that got there first. Nineteen sixty one, John Little published a proof relating how much work sits in a system, how fast it comes out, and how long a thing takes to get through. Work in the system equals throughput times lead time. A theorem, not a rule of thumb. So if you want shorter lead times and you aren't raising throughput, the only lever left is carrying less at once.

Miles: And the variability piece.

Nora: Also nineteen sixty one. John Kingman published an approximation for waiting time at a single server: three terms multiplied. Service time, utilization, variability. The utilization term is utilization over one minus utilization, so as you approach a hundred percent busy the wait doesn't climb gently. It runs away. And the variability term is built from how much arrivals and service times vary, so at identical utilization, more variation means a longer queue.

Miles: So the balanced plant.

Nora: Is one where every station sits at a hundred percent, which puts all of them on the runaway part of that curve with nothing in reserve. Goldratt put it in a novel in eighty four. Kingman had put it in a journal twenty three years earlier. Same statement.

Miles: And the five steps.

Nora: Identify the constraint. Decide how to exploit it, meaning get every usable minute out of it before you spend a cent. Subordinate everything else to that decision, which is the step nobody does. Elevate it, meaning add capacity. Then if the constraint has moved, go back to step one, and do not let inertia become the constraint.

Miles: Exploit before elevate.

Nora: Which is the most useful ordering in the book. Exploiting means the bottleneck doesn't stand idle over lunch, inspection moves in front of it so it never works on a part that's already scrap, and it only touches parts there's a customer waiting for. None of that costs money, all of it raises output, and buying a second machine is step four.

Miles: And batches.

Nora: Cut the batch and lead time falls, because a part spends most of its life waiting for the rest of its batch. He splits the batch you process from the batch you move, so a small quantity can travel ahead. Little's proof arriving as a shop floor instruction.

Miles: So where does it stop?

Nora: At the evidence, and I'd rather be precise than wave in its direction. The largest collection of results is a case survey by Steven Balderstone and Victoria Mabin, given to the New Zealand operational research conference in nineteen ninety eight and published in peer reviewed form in the International Journal of Operations and Production Management in two thousand three. Over three hundred published items, usable numbers on seventy seven companies.

Miles: And the numbers.

Nora: Large. Mean lead time reduction of sixty nine percent across thirty two observations. Mean inventory reduction of fifty percent across twenty eight. Due date performance up sixty percent across twelve. Revenue up sixty eight percent across eighteen, once you set aside one company that reported six hundred.

Miles: And the method behind the numbers.

Nora: Is what decides whether they're worth anything. These are firms that implemented it, wrote it up, and got published. No control group anywhere in the set. No randomization. Many of the before figures are reconstructed rather than measured, because plenty of these plants had no hard data to begin with. The authors say all of that themselves and name publication bias in their own limitations. Nobody writes up the plant where it failed.

Miles: Has anyone done it properly since?

Nora: In healthcare, partly. A systematic review in twenty twenty, in Health Systems, by Bacelar Silva, Cox and Rodrigues, gathered forty two implementations out of twenty nine publications, mostly hospitals, across nine countries. Mean waiting time down fifty percent, length of stay down thirty eight, overtime down ninety three.

Miles: Those are enormous.

Nora: And the sentence that goes next to them is that every implementation reported a positive outcome. Forty two out of forty two. That isn't a finding about the method. It's a finding about which implementations get written up, and the reviewers name it themselves.

Miles: Is there anything anywhere with a real comparison in it?

Nora: Simulation, and it's the most interesting result of the lot. Twenty seventeen, Thürer, Stevenson, Silva and Qu, International Journal of Production Economics. A simulated shop, seven stations, one of them the bottleneck, running Goldratt's release method against a family of load balancing methods across three levels of bottleneck severity and two shop layouts. Where the bottleneck was severe, Goldratt's method won. Where it wasn't, load balancing won, and their finding is that the balancing advantage stops mattering once a strong constraint exists.

Miles: So the book is true conditionally.

Nora: It's true where there's a dominant constraint, which is the case the novel is set in. It's weaker where capacity is genuinely spread about, which describes a lot of workplaces. And that's the tier statement, so I'll put it plainly. This is a novel. It's practitioner writing rather than peer reviewed research, and it was believed long before anybody tested it. What's unusual, and it's why this one comes through our segment in better condition than most, is that the arithmetic underneath had already been proved in journals by people with no connection to the book. Goldratt didn't demonstrate the mechanism. He dramatized one that was already established, then made claims about how far it generalizes that the record still hasn't settled.

Miles: Now our side of it.

Nora: And this is the one I'd want in a manager's hands, because the bottleneck frame is the kindest account of an uneven performer that management writing has produced.

Miles: Explain the kindness.

Nora: It takes the question off effort and puts it on placement. Picture somebody remarkable at one thing and slow at another. The ordinary reading is motivation, or attitude, or not caring about the parts they're weak at. Goldratt's reading is that a station is producing at capacity and the system has been arranged with that capacity in the wrong position. That's not a gentler verdict. It's a different question, and it has an answer somebody can act on.

Miles: And the arithmetic carries straight across?

Nora: Straight across, and harshly, in a useful direction. If the thing a person is slow at isn't what limits the team's output, improving it improves nothing. A development plan aimed at a non constraint is the mirage, in Goldratt's own terms: it costs that person enormously and returns no throughput. Meanwhile the thing they're extraordinary at, which is very often the actual constraint, is being interrupted by meetings.

Miles: Say the variability piece, because I think that's where this gets specific.

Nora: Goldratt's mechanism isn't slowness. It's fluctuation combined with dependency. And there's a meta analysis that matches that almost eerily. Michael Kofler and colleagues, twenty thirteen, Clinical Psychology Review, pooling three hundred and nineteen studies of reaction time variability in ADHD, children, adolescents and adults, against typically developing groups, corrected for measurement unreliability and for publication bias.

Miles: And the result.

Nora: An effect size around point seven six in children and adolescents and point four six in adults, on variability. And then the finding that does the work. Once you account for variability, the difference in average speed disappears. Once you account for average speed, the variability difference is still sitting there, and it's large.

Miles: So the difference isn't being slower.

Nora: It's being less predictable, on a laboratory measure, in aggregate. And that same review found adolescents and adults with ADHD indistinguishable from other clinical comparison groups on it, so it's nobody's fingerprint and a test for nothing. What it is, is exactly the quantity Kingman's formula takes as an input. More variability at the same utilization gives you a longer queue. Not a slower worker. A longer queue.

Miles: Which has a remedy in operations.

Nora: One that operations settled decades ago, and it isn't telling the machine to be steadier. It's protective capacity and a buffer in front of the constraint. Slack. And slack is the first thing an organization cuts, which gives you Jonah's balanced plant built out of people.

Miles: Say more.

Nora: One caution first. Karalunas and colleagues, twenty fourteen, Journal of Child Psychology and Psychiatry, meta analysed the same measure in autism and found it raised only when children with co occurring ADHD were in the sample. A warning against lumping, and I'd extend it to everything we're doing here. The uneven profile isn't one thing with one cause.

Miles: And is the unevenness measurable, or a figure of speech?

Nora: Measurable, carefully. Kuschner, Bennetto and Yost, two thousand seven, Journal of Autism and Developmental Disorders. Preschool children, three groups on a nonverbal test: autistic, non autism developmental delays, typically developing. The autistic group showed relative strengths in visual disembedding and detail focused work and relative weaknesses in abstraction, while both comparison groups came out roughly level. Small, very young, and it doesn't reach an adult at a desk. What it establishes is that unevenness is measurable rather than a way of talking.

Miles: And what does the record say about placement itself, since that's the move you're recommending?

Nora: Mixed, and I'd rather say so than oversell it. The big one is Kristof Brown, Zimmerman and Johnson, two thousand five, Personnel Psychology. A hundred and seventy two studies, eight hundred and thirty six effect sizes. Person job fit correlates with job satisfaction at about point five six across forty seven studies, with organizational commitment at point four seven, with intending to quit at minus point four six. With actual job performance, point two zero across nineteen studies, and the authors flag that one as the least dependable in the set, because the credibility interval around it includes zero.

Miles: So fit isn't a performance lever.

Nora: Not dependably, on that evidence, and most of it is people reporting their own sense of fit alongside their own attitudes, so it's self report at both ends. What fit predicts strongly is whether somebody's satisfied and whether they stay. Which is the point anyway. Putting an uneven performer where their capacity is the constraint isn't a productivity trick. It's the difference between keeping that person and losing them.

Miles: Fair enough.

Miles: One minute, in the frame we work in. The Goal says a system's output is set by one constraint, and improving anything else produces reports instead of results. If you lead people who think differently, or you're one of them, here's Monday. Find what's genuinely limiting the team's output before you write anybody a development plan, because a plan aimed at a non constraint costs that person everything and returns you nothing. Exploit before you elevate: protect the constraint's hours, and do the free things first, which means fewer interruptions and no rework from upstream. Expect variation rather than punishing it, and buy protection against it, because the answer to fluctuation has always been a buffer and never an instruction to be steadier. And watch for the balanced plant, everybody at a hundred percent with nothing in reserve. That one looks magnificent on a report and breaks on an ordinary Tuesday.

Miles: Now the part with the model in it.

Nora: Housekeeping first. The standing warnings about this whole category live past the music now, in a wording that doesn't move, so I'll leave them there and spend the time on this book.

Miles: Go on.

Nora: Step one is identify the constraint, and people get it wrong by asking. Ask a team where the holdup is and back comes the answer with the strongest feelings attached. Finding it is counting. Where things queue, how long they sit, which stage has a line that never clears. Tedious arithmetic over records you already keep, and tedious arithmetic just got cheap.

Miles: And exploiting it.

Nora: Is mostly taking work off the constraint, and much of what sits on a constrained person is not the constrained skill. The drafting, the summarizing, the formatting, the first pass nobody enjoys starting. Every minute handed back is throughput in the book's own sense, and it's step two rather than step four, because it doesn't cost you a hire. The transfer batch travels the same way: send a section the moment it's done instead of the document when it's whole.

Miles: And there's a switching cost in here.

Nora: There is, and it's been measured. Rubinstein, Meyer and Evans, two thousand one, Journal of Experimental Psychology, ran four laboratory experiments with adults either alternating between tasks or repeating one. Switching cost them time, the cost grew as the rules got more complicated, and it shrank when people were cued about what was coming. Small controlled studies with artificial tasks, so I wouldn't stretch it far. But protecting a constraint from being switched is a defensible instruction, and telling it in advance what's coming is a cheap one.

Miles: And the danger in it.

Nora: Is the sharpest thing this book has for anybody right now, and it's the robot from the cold open. These tools make non constraint work almost free. So a lot of teams are about to make far more of it, faster, with much better numbers attached, and ship exactly what they shipped last year. More drafts, more analyses, more decks, all queueing in front of the one person who has to decide. Local efficiency with a new engine bolted to it, which is the failure the novel opens on.

Miles: And the last word.

Nora: Goldratt's own. An hour saved where the constraint isn't is a mirage. Everything in this category will save you hours. Before you spend one, go and find the line.

Miles: That's The Goal.

Miles: Next time: The Phoenix Project, by Gene Kim, Kevin Behr and George Spafford, which takes this plot, keeps the impossible deadline, and sets the whole thing inside a company that ships software. Same Herbie, different building. The factory floor grew up and became the pipeline that gets code out the door.

Nora: Written by people who'd very obviously read this one twice. I'll be marking the correspondences.

Miles: Good work, Nora. Thanks.

Nora: Bring the next one.

Miles: Same reading room, tomorrow.

Miles: Sixty Books is a Stoar Consulting and Coaching production. Logan Williamson is a Licensed Professional Counselor in Texas and Louisiana. What happened here was teaching: one book, opened up for people who run teams and people who sit on them. Not coaching. Not counseling. And not a syllable of it was a judgment about you.

Nora: The citations sit on the episode page next to the transcript. Book thirty two arrives tomorrow.

Miles: Two declarations. The voices are generated. Miles isn't a person, Nora isn't a person, nobody here holds a license to do anything, and nothing said in it was aimed at whoever's listening. Every script gets drafted with AI in the room and then edited and approved by Logan Williamson, a Licensed Professional Counselor in Texas and Louisiana, at Stoar Consulting and Coaching.

Nora: And one note about the AI segment. The evidence that these tools help neurodivergent people at work is early. There are controlled trials now, but they measure things like conversation skill rather than how a job actually goes, and for work itself most of what exists is individual providers reporting success. So take it as promising and not proven. The tools also get rebuilt every few months, so whatever you set up today is built on something that moves.