« Making Advantage Sustainable | Main | Stats don't lie; but can they forget? »

December 5, 2005

Managing versus Measuring

We always say that we can't manage what we can't measure.

But what are you doing? Are you just measuring the management? Managing the measurement? Or, actually managing the item on which you're using the measurement and management processes?

I.

Management is essentially about determining whether the state of things needs adjusting and then intervening with adjustment actions that are appropriate. On their own, both environmental conditions and activities can be hugely indifferent to the particular desires of an organization, and to be routinely leveraged they must be perceived in consistent, non-random ways. Therefore, one can't manage what can't be described. Measurement is just one aspect of the fundamental management need for description, and it must be complemented by other forms. (See the illustration here and again near the end of this discussion.)

To understand this, it's important to immediately broaden our sensitivity about measurement -- away from "metrics", up to to the general functional problem of "description", and back down much closer to "definition".

Defining current states is a mandatory part of the management equation, but that doesn't amount to management. Instead, if the adjustments are the whole point of having management, then actually it's more critical that we define the adjustments. It's what kind of an adjustment, and how much of an adjustment, that drives the only difference we expect from what we'll later on still call "management"...

And just as all rectangles are not squares, not all definitions are measurements. Definitions are descriptions, and "measurements" are just a form of expression that descriptions can take.

But because measurements are so prolific, complicated and frequently debatable, we're concerned enough to ask the question, "can management be done without that form of description?"

II.

Let's say that you have a task to put a stone in a box, but the stone is initially too big -- which we determine simply from trial. How are you going to "manage to" get the stone into the box? Do you need measurement to get the stone certifiably "in the box"?

You could chip away at the stone to make it smaller, and keep trying after chipping it to put it in the box. Your action makes the stone different. Eventually you'll get the stone into the box.
- On the other hand, if you don't care how long it takes, how small the stone winds up, and other such things, you don't need measurement -- you just need the stone to finally fit into the box. But each attempt to fit the stone in the box is actually a measurement.
- And on the other hand, if it matters how long it takes, how big the stone finally is, etc., then of course you need even more measurement.

We tend to think of the latter case as being managed and the former not; but strictly speaking they are both managed. That is, in both cases, proactive adjustments are determined and made deliberately towards the target state until the target state is true. We see that measurement figures in differently but both times. Most importantly, the target state is true only when we agree that it is true, and we use the measurement to gauge the distance from the agreement.

So far, to act towards the target we still haven't necessarily invoked measurement, but to know that we're reaching the target we must have it.

Practically speaking, the key element is actually the agreement beforehand, which means that the terms of agreement are the key to describing the target state. These terms may be non-metrical, while still being highly differentiating. For example, objectives are often stated to represent the terms of agreement by which most comparisons will be made. Compatibility, not compliance, is usually the subject of this perspective on things.

But let's get back to measurement. Measurement provides a kind of additional certainty to definitions, by sharpening the definition's ability to express distinction -- typically in terms of "amounts". For example, even a qualitative binary distinction - true vs. false - can be considered almost purely measurement if the real question posed is "how much 'false' is there?" and the answer is "None."

Yet in that example, it's obvious that we must still know beforehand which conditions actually mean "True". Can we identify (define) the conditions without measurement?

Sure, if we can distinguish them in other ways.

First we should look at why we can do that, and then at why we should do that.

III.

In the example above, it can be argued that management didn't require measurement to make "stone in the box" become true. But measurement is required to confirm that it actually is finally true. So the point is that the success of the management is what we need measurement to expose.

Let's make that thought more concrete with another example. If a test case is looking for exactly a 3-foot length, any length tested will need to have been measured before we can say that it complies. Compliance is determined purely by comparison to some standard (in this case, something that we already know is 3 feet long). What makes the comparison work is that the standard has already been selected. Actually doing the comparison -- that is, simply applying the standard to the test item -- is an action that might be "managed" or accidental, but we can confirm the outcome only through measurement, and the confirmation is what tells us whether or not we need to try againto hit our target.

This is all adding up in a certain way:
- definition is aimed at identifying the distinction of the target conditions
- the difference is described and recognizable by terms of agreement
- compliance to those terms is largely detectible or provable through testing
- the testing would rely on a standard to help establish the degree of compliance to the terms
- more than one condition might be part of the standard

But there is another way to work in terms of a standard: procedure.

Procedure prescribes steps that are expected to produce compliance. The assumption built into a procedure is that during execution each step of the procedure should become true. Largely without measurement, we can simply follow the procedure. In this case, most measurement used is dedicated to determining how much execution complies with the procedure (supervision), and to confirming the effects of making adjustments (intervention) -- but the choice to use the procedure is far more significant: it means that given the responsibility to direct the course of events, we have agreed to the procedure's logic. Exercizing the logic, with whatever level of competence, is the most differentiating characteristic of what we are calling "management"...

IV.

So, as it turns out, the urgency about measurements is not really about manageability but instead is about predicting the effectiveness of the management, and about increasing the accountability of the management. In short, they are about making management scientific.

However, management need not be scientific in order to be management. What management really depends on even more than measurement is logic. Measurement should be a means of testing and tracking the logic, but logic must be derived even in the absence of desired measurements. This takes place in the form of assumptions and objectives. Management logic can form a closed loop or system when its assumptions and objectives are followed with ratings. (Current ratings influence future assumptions.)

In completing that loop we also get the broad outline of the framework for management description:
- Assumptions represent the key initial distinctions acknowledged in the management effort; this includes discovery of identities and definitions
- Objectives represent the point of view that finds comparative significance in what is done and monitored; this includes test criteria and terms of agreement
- Ratings represent the actual visibility on the value of conditions, achievable from foresight to hindsight; this includes standards and priorities.

With those anchoring observations, we can see the universe of description that management needs.

This framework intends to position the many modes and artifacts of description in a way that highlights where they make the key difference to management's overview of conditions. Against that, it sorts out and addresses the basic management concerns that drive decisions to intervene:
- preferred states;
- adjustment progress; and,
- the impact of changes.

In the illustration it is easy to see the range of formats for describing conditions, along with the fact that they are not necesarily metrical in nature. However, it is also evident that measurement plays some role in every area of description -- such as, by establishing trending ("track"), marking off milestones ("score"), or detailing cost/time limitations on requirements ("specify")...

V.

Our key management challenge is to conceptually arrive at definitions identifying one thing versus another, so that we can properly identify their relationships and leverage the relationships.

These definitions, when treated *as if* they are 'facts', are the normal basis of managerial logic. Communities of practice may develop and even enforce 'standards' of definition, but this ultimately does not prevent management from being practiced based on other standards or knowledges and in non-scientific yet still logical modes.

As for measurement, it has the task of identifying and comparing differences in an accounting mode. Three other issues ensue:
- to actually do the measuring;
- to measure the things that matter; and,
- to match the right measures with the right management.

Otherwise, measures simply become an unreliable tool being used for "solving the wrong problem."

Posted by Malcolm Ryder at December 5, 2005 7:46 AM

Trackback Pings

TrackBack URL for this entry:
http://www.malcolmryder.com/cgi-bin/mt-tb.cgi/178

Comments

Post a comment