|
A Guide to Implementing the Theory of Constraints (TOC) |
|||||
Evaluating Change When we set out to
implement change we must remember that there are 3 possible outcomes. These outcomes are; (1) A change which is a
significant improvement. (2) A change which is neither
a significant improvement nor a significant decline. (3) A change which is a
significant decline. Naturally enough, it is
the first option that we are really seeking.
We want to make a difference, and we want that difference to be
manifestly positive. In order to do
so, we must make decisions prior to carrying out the desired actions, and to
be certain in the knowledge that those decisions will deliver the necessary
results that we seek. How we evaluate the
improvement will depend upon the goal of the system. If the goal is a monetary one, then the
evaluation is relatively straightforward.
And that is what we will concentrate on here. In not-for-profit, or more correctly,
for-cause situations how we evaluate an improvement is a little more
involved; however, if you look at the argument for healthcare (supply chain
section) then you will find some good indications of how this can be
achieved. We are evaluating changes
within the context of the whole system – or the system as a whole. We are not interested in local improvements
that do not have system-wide impact. How, then, would we judge an impact in such
a circumstance? We need a
context. We already have one, let’s
revisit it. On the measurements page
we derived our rules of engagement.
These tell us how to define the entity that we want to improve. We define the boundaries, the goal, the
necessary conditions, and the fundamental measurements. Without these, we do not have a context
within which to evaluate change.
Moreover, this forces us to determine what it is that constrains us
from moving towards our goal; we have to define the role of the constraints. The constraints are
central to our ability to move forward.
In order to define the role of the constraints we need to invoke our
plan of attack, the one we developed on the process of change page. Of course, our plan of attack is Goldratt’s
focusing process. The second step of
this plan, where we decide how to exploit the constraints, is the step that
provides commonality between these two schemes. We have previously
summarized the relationship between the rules of engagement and the plan of
attack as follows;
In order for a change to
be an improvement it must either have a direct positive effect upon the
current exploitation or elevation of the system’s constraints, or an indirect
effect via improved subordination which in-turn ought to improve the
exploitation or elevation, either now or in the future. To quantify these effects
we must return to our fundamental measurements. In the first page, the page
on measurements, we briefly introduced the concepts of; throughput, inventory/investment,
and operating expense; a
triumvirate set of measures for quantifying effects in Theory of Constraints. Throughput as you may
remember was described as; Throughput =
Sales - Totally Variable Costs From this we came to
define our net profit as; Net
Profit = Throughput - Operating Expense And return-on-investment
is;
The reason that we can do
so much with so little is because of the fundamental relationships that exist
between each measure. They are systemic. Let’s try to reinforce
the fundamental and systemic nature of these measures by way of analogy. By this means we will be in a far stronger
position to understand change and how to evaluate it. The analogy is a see-saw. A see-saw!
How does the evaluation of change relate to a see-saw? Well, let’s have a look. Let’s draw a simple see-saw as a start.
The two equal masses – “people” are located
equidistant from the mid-point of the plank, so let’s label that.
Let’s see.
Can we use this simple analogy of a see-saw for
evaluating internal management decisions, change in other words? In terms of physical aspects it is apparent that we seek to leverage inputs
of some kind via a process of some sort in order to produce outputs. In fact, the process does not exist in
isolation but rather it exists in conjunction with a set of operating
assumptions; the things that we call policies. How then would this look using our
model? Let’s see.
How then would our model look in terms of financial
aspects? In the terms of financial aspects it is apparent that
we seek to leverage expenditure via investment to produce income. Once again the investment does not exist in
isolation but rather it exists in conjunction with a set of working
assumptions; policies once again.
Let’s see how this looks.
We need to ask then; will this simple analogy, a
see-saw, also work as a description for evaluating change in Theory of
Constraints? Well, I think so, so
let’s try.
It seems then, that our analogy will hold for our
fundamental measures. Great. Any change in throughput, or operating expense may
change the balance of our system. Do
you agree? Our analogy shows the
interrelationships between these various aspects. Do you want to push the analogy a little bit
further? What is our profit then? Let’s have a look.
What about the balance point then?
We know the location of the balance point, but this
begs a question. What is the fulcrum
that we leverage across? Let’s have a look.
I know that all too often we loosely talk about
leveraging the constraint – we have used that language throughout these
webpages and it is probably ingrained.
But in reality we are leveraging our entire system over the fulcrum
– time – and the only way that we can do that, either literally or
metaphorically, is via the constraint.
So we leverage the system via the constraint for a given unit
of time. So, yet another question; what exactly is the
constraint in our analogy then?
Now that we have identified the constraint, how can we get more of this limiting
factor? How in our metaphor can we get
more people sitting balanced on the right-hand side? How can we improve the Throughput? How can we improve the profit? There are two answers to these questions,
and they are that we can increase the productivity, and/or we can increase
the production. We need to tease these
two strands apart in order to better understand each of them. Let’s do that. Making a distinction between productivity and
production is important in understanding how to most effectively drive
improvement, and such a distinction is also useful in developing our
understanding of the dynamics of exploitation, subordination, and
elevation. Production is the simpler,
and certainly more familiar of the two concepts, so let’s start with that. Essentially any increase in production is a pro rata
increase in both inputs (operating expense), and outputs (Throughput). Let’s investigate this with our see-saw
analogy. Let’s start again with our original model with a
balance point located 3/4 of the way along the plank.
Increasing production, seductive as it is – after
all this is what almost everybody else does – is nowhere near as sexy as
improving productivity per se. Moreover,
if we were to go around doing what everyone else does then there is hardly
any strategic advantage to be had at all. So let’s investigate the impact of improving
productivity; many people talk about increasing productivity but few actually
manage to do it. Doing it is not at
all difficult if we have focus. Rather than settling for a pro rata increase in both
operating expense and throughput, which means constant productivity – only
more of it, we actively seek to decouple throughput from operating expense,
which in-turn means increased productivity.
Throughput should increase and ideally operating expense should remain
static or even decrease; something other than additional operating expense
drives the additional throughput. It
is the leveraging of the entire system via the constraint’s throughput
relative to the fulcrum, time, that drives the additional throughput. Let’s show this by example. Let’s start again with our original model with a 3:1
ratio.
An increase in productivity will in-turn
substantially increase profit. Let’s
have a look at that.
We obtain better leverage by better exploitation of the constraint (the
secondary plank becomes longer) and by better subordination of the non-constraints. Often the simplest way to obtain an
increase in leverage is to remove or modify some current policy. Organizations abound with policy; that is,
after all, one way in which to standardize matters, and without
standardization there can be no base from which to improve. But what if the standardization causes us
to stagnate instead of improve? Policy
also allows us to react quickly without reinventing the wheel each time. But what if we no longer need a particular
reaction and yet we still have the policy?
Removal of outdated or inappropriate policy unblocks access to current
capacity and increases productivity. Now, if we are still bored with our newfound
increase in productivity, then we can still increase our production after we
have increased our productivity – given that our capacity allows for it. It pays in more ways than one to increase
relative productivity first, and then absolute production second, rather than
the other way around. Always aim for
capability before capacity. That is why the 5 focusing steps; our plan of attack
goes; identify, exploit/subordinate,
elevate – in that order. Most
firms go; identify, elevate – every time.
In fact that is unfair, most firms miss the identification stage and
have a scatter gun approach of; elevate, elevate, elevate. Hardly a wise use of cash, and a total
absence of any systematic decision analysis. In reality, often both productivity and production
are inexorably mixed together, but we need to understand the dynamics of each
component if we are to better understand how to correctly influence the whole
– even if later on we can’t so neatly break the whole back into constituent
parts as we have here. It is apparent from the logic of this discussion
that as the lever moves with respect to the fulcrum; the productivity,
throughput, and hence profit, should trend towards infinity. But we are getting ahead of ourselves. None of us are making infinite profits yet
(or if we are, then we certainly haven’t told Inland Revenue about it). So, this begs a question. Why aren’t we making infinite profits yet? Well, a valid reason might be a finite
capacity or capability of the current constraint; we are unable to move the
balance point any closer to the end of the lever. We can neither exploit the constraint nor
subordinate the system any further, even though the demand is there? What shall we do? Well, why don’t we make the lever even longer? Let’s have a look at a new aspect;
investment. In our analogy additional investment means that our
lever becomes a little longer. The
effect of the investment in this instance is to both exploit & elevate the existing constraint or to better
subordinate the non-constraints which in-turn exploits the constraint. Let’s work from our current state where we
have 7/8’s of the lever on one side and 1/8 on the other. Here is our starting point.
The effect of the investment is to increase the
physical leveragability of the system even though the absolute position of
the balance point remains static.
Effectively we have increased the productivity of the system by
capital investment. This is
interesting (to me). Here we have both
elevation (cash from outside the
system was brought inside – even though it is not an increase in operating
expense) and exploitation (the
absolute position of the balance point did not change, but the position
relative to the whole plank did
change). Alright, maybe that is
pushing our metaphor about as far as it should go at the moment. Let’s now return to the formulae that express the
reality of these simple diagrams to further evaluate the situation. We introduced 3 equations in the section on
fundamental measurements, let’s repeat two of them here, one for throughput
and one for profit (or operating surplus); Throughput =
Sales - Totally Variable Costs and Net
Profit = Throughput - Operating Expense Of course we can combine these into one statement Net
Profit = Sales - Totally Variable Costs - Operating
Expense However, let’s confine ourselves to the simpler
version Net
Profit = Throughput - Operating Expense And let’s compare this directly with the simplest of
our see-saw analogies. Here is the
analogy.
Net Profit =
3 Units of Throughput - 1 Unit of Operating Expense = 2 Units Just as we drew it,
We can see that the fulcrum is represented in the
diagrams and we can see that its positioning under the balance point is
critical, and we know that it represents time, yet it seems to disappear from
our equations. Let’s clarify this
issue. The fulcrum is time – the one thing we don’t seem to
be able to generate any additional quantity of, and time is present in our
equations, but we seem to have been a little lax in making it explicit. Really our equations should read as
follows. Throughput should be;
And one again, if we combine these equations we get;
We are more interested in evaluating change before
we make the change. We want to know
the outcome of a potential decision before we take action to implement it as
an actual decision. And for this we
need some critical information. We need to be able to determine the Throughput
through a unit of constraint capacity in relation to time.
This is the
major decision analysis that we make.
We need to examine this in detail. Let’s return to our original case for a moment.
Upon the improved leverage of the example above we
found the following;
But equally, we might also have found this;
If we substituted children for all of the adults we
could even have found this;
How can we know ahead of time what the outcome of
these types of substitutions will be?
Graphically it seem obvious, we need to know the output value, the
weight, for each type of output in this system relative to the unit constraint capacity. Let’s show this.
It is a simple step to move our analogy from output
to throughput so that we can evaluate the financial aspects. Let’s
have a look.
We have produced a normalized throughput per unit of
output as viewed from the perspective of the constraint. But,
we have nearly lost sight of our fulcrum (again). What has happened to our measure of time? Mention was made of the short-hand expression “T/cu”
or Throughput per constraint unit earlier.
This short-hand is partially responsible for the apparent lack of time. It is there, however. The full expression should be “throughput
per constraint unit per unit time,” or T/cu/t. In our simple analogy our constraint unit is
seating, and thus we would evaluate Throughput as Throughput $ per seat per
ride. “Seat” is the constraint unit,
“ride” is the expression for time. Let’s look at a few other general cases. What about a sunshine factory? And by that I mean an outdoor agricultural
or horticultural enterprise. The
constraint unit here is available productive area, and the decision analysis
becomes Throughput $ per acre or hectare per season or per year (T$/hectare/year). What about an indoor retail operation? Something that doesn’t make anything; just
buys and sells. The constraint unit
here is again productive area, if might be square meters or square feet of
floor space, or square meters or square feet of shelf space if there is a
vertical component as well, and the decision analysis becomes Throughput $
per square meter per week or per month depending on the rate of turnover (T$/m2/week). Supermarkets tend to use linear meters of
“facing” assuming that we buy in proportion to what we see. If the facing all has the same volume
stacked behind it, then there would seem to be little difference in the
various units. Larger items in a sales system where a sale is
concluded after a sales process, then the constraint should be the number of
contact sales hours that the sales people have. The decision analysis becomes Throughput $ per
sales person per hour or day (T$/sales
person/hour). In manufacturing the constraint is most usually a
machine or group of machines and this is the constraint unit, the unit of
time is most commonly minutes because manufacturing steps are more commonly
completed within minutes or hours rather than days. The throughput decision analysis becomes
Throughput $ per machine per minute (T$/machine/minute). Some examples that I know of are
people-paced rather than machine-paced and the throughput decision analysis
becomes Throughput $ per man per hour (T$/man/hour). What then of projects? The constraint is the number of resources
working on the critical chain. The
Throughput decision analysis becomes Throughput $ per critical chain person per
project week or month (T$/critical
chain person/month). Remember these are decision analyses; the analysis
of various choices before we embark
on a decision. Once we have made a
commitment to the customer we can’t internally re-prioritize according these
values. With this new information under our belt, now, at
least, we can predict the Throughput for the following case before we
actually do it.
What about the other case?
Graphically this is just plain obvious, we can see,
and we know from our own direct experience with see-saws. But trust me, in most organizational
systems this is anything but clear, and one good reason for this is that
almost no one in most organizations has ever considered this before. “We have to evaluate the impact, not of a product,
but of a decision. This evaluation
must be done through the impact on the system’s constraints. That’s why identifying the constraints is
always the first step (1).” We have to
know where the constraint, is; and we have
to know the Throughput value of the output with respect to that
constraint. So, anyone
can work out the Throughput retrospectively for any period. No
one can evaluate the Throughput proactively for the current or future
periods without explicit knowledge of the Throughput value generated with
respect to the constraint. In some
instances this might be quite obvious; most often, however, it is not. And when it is done there are most often
some surprises in the relative ranking of the outputs. This brings us to a very important point. We all know from our own personal experiences with
see-saws, that if we change just one important thing then the whole balance
may change. If we move the plank a
little, or if someone gets on, or if someone gets off, or even it someone
changes ends, then, so too, does the balance.
And so too, with our system under investigation. We didn’t know previously that once we elevated this
system that children might get on, or that adults might get off. But every time we prepare to change the
constraint that is exactly what we must evaluate for. We must evaluate for the new mix that could
arise. If we think that we will
elevate a constraint to the extent that we will break it (and thus a new
constraint presents itself) then the unit throughput values, and thus the
individual ranking, may also change and therefore maybe also our tactics for
exploitation will change as well. We must predict the outcome ahead of implementing
the decision. “You see, in the ‘cost
world’ almost everything is important, thus changing one or two things
doesn’t change the total picture much.
But this is not the case in the ‘throughput world.’ Here, very few things are really
important. Change one important thing
and you must re-evaluate the entire situation (2).” In internally constrained systems we can not satisfy
market demand. We can show this with
our analogy? Of course we can. The beam is full, we could get other people
on, if only they would fit.
We have, loosely speaking, begun to evaluate
decisions about the composition of the physical output – the so-called
production mix. Now, there is almost
nothing, either positive or negative, that a good production manager can not
ascribe, in one way or another, to the changes in the production mix. The production manager can do this without
fear of contradiction because, in fact, almost no one else understands the
true impact of the production mix – often not even the production manager! But we do understand – only too well. We do, because we know how to strip out all of the
raw material costs or 1:1 variable costs in the production mix leaving
us with the bare essentials – we could call this the throughput mix
(3, 4). Moreover, we know that we must
evaluate the throughput mix in relation to one thing and one thing only, the
amount of resource consumed to produce the throughput mix on the constraint. Let’s therefore look at this in a little more
detail. We need to better delineate
some aspects that are particularly important in internally constrained
environments. Maybe we should describe
these as generic tactics. We need to do this in order to later
appreciate some of the subtle changes that occur in the tactics once a system
becomes externally constrained. It becomes part of the exploitation strategy of
internally constrained systems to maximize the throughput mix by including,
as much as possible, products that generate high throughput per unit time on
the constraint; these are the adults of our analogy. In shorthand we might describe these as
high “T/cu” products, where “T” means throughput and “cu” means constraint
unit. We saw this aspect demonstrated
so well in the P
& Q analysis. However, I want to introduce the word “grade” to
describe this aspect of throughput mix.
We want to substitute, wherever possible, higher grade unit throughput
for lower grade unit throughput. We
want to produce “stars” not “dogs” if the market will allow us – and it
should. The overall grade is a
reflection of the average throughput per unit. Moreover, let’s call the current market capacity
“market volume” and the profit “cash flow.”
Let’s have a look at the relationships.
And I would like to embellish this further – or at
least make it easier for me to understand – by adding some “gauges” to this;
a sort of metaphorical dashboard for our system. And let’s assume for the moment that this
is the view prior to implementing our new-found knowledge.
Therefore, let’s exploit this internally constrained
system and see what happens.
|
