The Goal: a process of Ongoing Improvement

less than
the
demand placed upon it. And a non-bottleneck is any resource whose capacity
is greater than the demand placed on it. Got that?’’

---

"Right,’’ I tell him.
"Once you have recognized these two types of resources,’’ says Jonah, "you
will begin to see vast implications.’’
"But, Jonah, where does market demand come in?’’ Stacey asks. "There has
to be some relationship between demand and capacity.’’
He says, "Yes, but as you already know, you should not balance 
capacity
with demand. What you need to do instead is balance the 
flow of product
through the plant with demand from the market. This, in fact, is the first of
nine rules that express the relationships between bottlenecks and non-
bottlenecks and how you should manage your plant. So let me repeat it for
you: Balance flow, not capacity.’’
Stacey is still puzzled. She says, "I’m not sure I understand. Where do the
bottlenecks and non-bottlenecks come into the picture?’’
Jonah says, "Let me ask you: which of the two types of resources determines
the effective capacity of the plant?’’
"It would have to be the bottleneck,’’ she says.
I say, "That’s right. It’s like the kid on that hike last weekend —Herbie. He
had the least capacity and he was the one who actually determined how fast
the troop as a whole could move.’’
"So where should you balance the floor?’’ asks Jonah.
"Oh, I see,’’ says Stacey. "The idea is to make the flow through the
bottleneck equal to demand from the market.’’
"Basically, yes, you’ve got it,’’ says Jonah. "Actually, the flow should be a
tiny bit less than the demand.’’
"How come?’’ asks Lou.

---

"Because if you keep it equal to demand and the market demand goes down,
you’ll lose money,’’ says Jonah. "But that’s a fine point. Speaking
fundamentally, the bottleneck flow should be on a par with demand.’’
Bob Donovan is now making various noises, trying to get into the
conversation.
"Excuse me, but I thought bottlenecks were bad,’’ says Bob. "They ought to
be eliminated where possible, right?’’ "No, bottlenecks are not necessarily
bad—or good,’’ says Jonah, "they are simply a reality. What I am suggesting
is that where they exist, you must then use them to control the flow through
the system and into the market.’’
That makes sense to me as I’m listening, because I’m remembering how I
used Herbie to control the troop during the hike.
"Now I have to run,’’ says Jonah, "because you caught me during a ten-
minute break in a presentation.’’
I jump in. "Jonah, before you go—!’’
"Yes?’’
"What’s our next step?’’
He says, "Well, first of all, does your plant have any bottlenecks?’’
"We don’t know,’’ I tell him.
"Then that’s your next step,’’ he says. "You have to find this out, because it
makes an enormous difference in how you manage your resources.’’
"How do we find the bottlenecks?’’ says Stacey.
"It’s very simple, but it would take a few minutes to explain. Look, try to
figure that out for yourselves,’’ says Jonah. "It’s really easy to do if you think
about it first.’’

---

I say, "Okay, but....’’
"Good-bye for now,’’ he says. "Call me when you know if you have a
bottleneck.’’
The speaker phone issues a click, followed by a fuzzy hum.
"Well... what now?’’ asks Lou.
"I guess we look at all our resources,’’ I say, "and compare them against
market demand. If we find one in which demand is greater than capacity, then
we’ll know we’ve got a bottleneck.’’
"What happens if we find one?’’ asks Stacey.
"I guess the best thing to do would be what I did to the scout troop,’’ I say.
"We adjust capacity so the bottleneck is at the front of production.’’
"My question,’’ Lou says, "is what happens if our resource with the least
capacity in fact has a capacity greater than what market demand calls for?’’
"Then I guess we’d have something like a bottle without a neck,’’ I say.
"But there would still be limits,’’ says Stacey. "The bottle would still have
walls. But they’d be greater than the market demand.’’
"And if that’s the case?’’ asks Lou.
"I don’t know,’’ I tell him. "I guess the first thing to do is find out if we’ve
got a bottleneck.’’
"So we go look for Herbie,’’ says Ralph. "If he’s out there.’’
"Yeah, quick, before we talk ourselves to death,’’ says Bob.
I walk into the conference room a few days later and there’s paper

---

everywhere. The main table is covered with computer print-outs and binders.
Over in the corner, a data terminal has been installed; next to it, a printer is
churning out even more paper. The wastebaskets are full. So are all the
ashtrays. The litter of white styrofoam coffee cups, empty sugar packets and
creamer containers, napkins, candy bar and cracker wrappers, and so on is
scattered about. What has happened is the place has been turned into our
headquarters in the search for Herbie. We have not found him yet. And we’re
getting tired.
Sitting at the far end of the main table is Ralph Nakamura. He and his
data processing people, and the system data base they manage, are essential
to the search.
Ralph does not look happy as I come in. He’s running his skinny fingers
through his thinning black hair.
"This isn’t the way it’s supposed to be,’’ he’s saying to Stacey and Bob.
"Ahh, perfect timing,’’ says Ralph when he sees me. "Do you know what we
just did?’’
"You found Herbie?’’ I say.
Ralph says, "No, we just spent two and a half hours calculating the demand
for machines that don’t exist.’’
"Why’d you do that?’’
Ralph starts to sputter. Then Bob stops him.
"Wait, wait, wait a minute. Let me explain,’’ says Bob. "What happened was
they came across some routings which still listed some of the old milling
machines as being part of the processing. We don’t use them—’’
"Not only don’t we use them, just found out we sold them a year ago,’’ says
Ralph.

---

"Everybody down in that department knows those machines aren’t there
anymore, so it’s never been a problem,’’ says Bob.
So it goes. We’re trying to calculate demand for every resource, every piece
of equipment, in the plant. Jonah had said a bottleneck is any resource which
is equal to or less than the market demand placed on it. To find out if we’ve
got one then, we concluded we first would have to know the total market
demand for products coming out of this plant. And, second, we would have to
find out how much time each resource has to contribute toward filling the
demand. If the number of available hours for production (discounting
maintenance time for machines, lunch and breaks for people, and so on) for
the resource is equal to or less than the hours demanded, then we know we’ve
found our Herbie.
Getting a fix on the total market demand is a matter of pulling together data
which we have on hand anyway—the existing backlog of customer orders,
and the forecast for new product and spare parts. It’s the complete product
mix for the entire plant, including what we "sell’’ to other plants and
divisions in the company.
Having done that, we’re now in the process of calculating the hours each
"work center’’ has to contribute. We’re defining a work center as any group
of the same resources. Ten welders with the same skills constitute a work
center. Four identical machines constitute another. The four machinists who
set up and run the machines are still another, and so on. Dividing the total of
work center hours needed, by the number of resources in it, gives us the
relative effort per resource, a standard we can use for comparison.
Yesterday, for instance, we found the demand for injection molding machines
is about 260 hours a month for all the injection molded parts that they have to
process. The available time for those machines is about 280 hours per month,
per resource. So that means we still have reserve capacity on those machines.
But the more we get into this, the more we’re finding that the accuracy of our
data is less than perfect. We’re coming up with bills of material that don’t
match the routings, routings that don’t have the current run-times—or the
correct machines, as we just found out—and so on.

---

"The problem is, we’ve been under the gun so much that a lot of the updating
has just fallen by the wayside,’’ says Stacey. "Hell, with engineering changes,
shifting labor around, and all that happening all the time, it’s just plain tough
to keep up with it no matter what,’’ says Bob.
Ralph shakes his head. "To double-check and update every piece of data
relevant to this plant could take months!’’ "Or years,’’ mumbles Bob.
I sit down and close my eyes for a second. When I open my eyes, they’re all
looking at me.
"Obviously, we’re not going to have time for that,’’ I say. "We’ve only got
ten weeks now to make something happen before Peach blows the whistle. I
know we’re on the right track, but we’re still just limping along here. We’ve
got to accept the fact we’re not going to have perfect data to work with.’’
Ralph says, "Then I have to remind you of the old data processing aphorism:
Garbage in, garbage out.’’
"Wait a minute,’’ I say. "Maybe we’re being a little too methodical.
Searching a data base isn’t the only way to find answers. Can’t we come up
with some other faster way to isolate the bottleneck—or at least identify the
candidates? When I think back to the model of the boys on the hike, it was
obvious who the slower kids were on the trail. Doesn’t anybody have any
hunches where the Herbie might be in the plant?’’
"But we don’t even know if we’ve got one yet,’’ says Stacey.
Bob has his hands on his hips. His mouth is half open as if he might say
something. Finally, he does.
"Hell, I’ve been at this plant for more than twenty years. After that much
time, I know where the problems usually seem to start,’’ he says. "I think I
could put together a list of areas where we might be short on capacity; at least
that would narrow the focus for us. It might save some time.’’
Stacey turns to him. "You know, you just gave me an idea. If we talk to the

---

expeditors. They could probably tell us which parts they’re missing most of
the time, and in which departments they usually go to look for them.’’
"What good is that going to do?’’ asks Ralph.
"The parts most frequently in short supply are probably the ones that would
pass through a bottleneck,’’ she says. "And the department where the
expeditors go to look for them is probably where we’ll find our Herbie.’’
I sit up in my seat. "Yeah, that makes a lot of sense.’’
I stand up and start to pace.
"And I’ll tell you something 
I
just thought of,’’ I say. "Out on the trail, you
could tell the slower kids by the gaps in the line. The slower the kid, the
greater the distance between him and the kid in front of him. In terms of the
analogy, those gaps were inventory.’’
Bob, Ralph, and Stacey stare at me.
"Don’t you see?’’ I ask them. "If we’ve got a Herbie, it’s probably going to
have a huge pile of work-in-process sitting in front of it.’’
"Yeah, but we got huge piles all over the place out there,’’ says Bob.
"Then we find the biggest one,’’ I say.
"Right! That’s got to be another sure sign,’’ says Stacey.
I turn and ask, "What do you think, Ralph?’’
"Well, it all sounds worth a try,’’ says Ralph. "Once you’ve narrowed the
field to maybe three of four work centers, it won’t take long for us to check
your findings against the historical data just to be sure.’’
Bob looks at Ralph and says in a kidding voice, "Yeah, well, we’ve all seen
how good that is.’’

---

But Ralph doesn’t take it in a kidding way. He looks embarrassed.
"Hey, I can only work with what I’ve got,’’ he says. "What do you want me
to do?’’
"Okay, the important thing is that we have new methods to try,’’ I say. "Let’s
not waste time pinning the blame on bad data. Let’s get to work.’’
Fueled by the energy of new ideas, we go to work, and the search goes
quickly...so quickly, in fact, that what we discover makes me feel as though
we’ve run ourselves straight into a wall.
"This is it. Hello, Herbie,’’ says Bob.
In front of us is the NCX-10.
"Are you sure this is a bottleneck?’’ I ask.
"There’s some of the proof,’’ he says as he points to the
stacks of work-in-process inventory nearby—weeks of backlog according
to the report Ralph and Stacey put together and which we reviewed about an
hour ago.
"We talked to the expeditors,’’ says Bob. "They say we’re always waiting
for parts from this machine. Supervisors say the same. And the guy who runs
this area got himself a set of earplugs to keep him from going deaf from all
the bitching he gets from everyone.’’
"But this is supposed to be one of our most efficient pieces of equipment,’’ I
say.
"It is,’’ says Bob. "It’s the lowest-cost, highest-rate means we have of
producing these particular parts.’’
"So why is this a bottleneck?’’

---

"This is the only one like it we’ve got,’’ he says.
"Yes, I know that,’’ I say, and I stare at him until he explains.
"See, this machine here is only about two years old. Before we installed it, we
used other machines to do what it does. But this machine can do all the
operations that used to take three different machines,’’ says Bob.
He tells me about how they used to process these parts using the three
separate types of machines. In one typical instance, the process times per part
were something like two minutes on the first machine, eight minutes on the
second, and four minutes on the third—a grand total of fourteen minutes per