This is a reply to my message on measurements from a close, personal
friend. He and I spent over three years working together at Novell. He and
I had breakfast nearly every morning to discuss the world around us. I
thought his thoughts were pertinent.
Ben
[Host's Note: Below is Ben's friend's msg. ...Rick]
On Tuesday, October 28, 1997 5:57 AM, Benjamin B. Compton wrote:
> There are three things we might consider when talking about measurements
> and measuring:
>
> 1- Businesses operate like Complex Adaptive Systems (cas). And one of
> the
> properties of a cas is nonlinearity (see Holland, "Hidden Order"). Our
> traditional measurement tools are excellent at describing the linear
> properties of a system. They don't do so well at describing the
> nonlinear properties.
For instance? If we can't measure it, how did we know it was non-linear? I
tend to subscribe to the school that says anything real can be measured,
including, yes, love.
> 2- The Heisenberg Uncertainty Principle certainly has an affect on our
> organizations: By measuring them we change them. Ultimately we're left
I think this is a nice metaphor, but a separate issue from quantum
uncertainty. Do you mean this literally? As in, if we measure exactly how
far a junior executive has his head up a vice president's butt that we
can't tell the precise velocity with which it was inserted?
> a measurement of the past, and that's about it. Outside of dialogue, I
> know
> of no way to measure the "present moment" or it's capacity to create a
> desired future. Intuition seems to be our best tool for grasping the
> "hear
> and now."
If you say, now is better than then, you have made an ordinal measurement.
I think most measurement schemes in business fail because of the great
difficulty in gathering data. It's not that you don't know how to tell
worse from better, but that you don't have time to observe.You try to find
an operational indicator to stand for what you really want to measure
because its impractical to measure directly.
For example, I was a big advocate of rewarding people for writing TIDs at
Novell. I believed that number of TIDs written would reasonably closely
track real knowledge generated. As it turned out, when you made the sad
sacks who didn't care about contributing to the knowledge base meet those
numbers, all they did was spew noise into the databases. It wasn't that I
couldn't measure what was good and what was crap, just that there's no
time to directly observe and measure that kind of performance.
The number of TIDs written was a bad thing to reward people for because
people just conned the system. When you measure, you have to take into
account that people will be productive--they will try to produce what
gives them the most reward with the least effort. They are efficient,
according to their interests and values. You can counteract their tendency
to finesse the system with a web of balancing measures, but that gets to
be as onerous as directly measuring what you want from people.
I think measurement is critical for feedback purposes, for getting a
reality check. You measure after you predict, or else you are just
admitting you don't really believe in your own predictions. But
measurement is poorly suited to enforcing discipline or eliciting
commitment or creativity. In business, I think it makes more sense to
follow your hunches, go on your gut feelings about what's really going on,
but then predict that if you are right, then this measurable x should
happen. If x doesn't happen, you don't make excuses or talk about
unforseeable factors, you admit you were wrong. If you were right, you
would have forseen those factors.
I think one reason Novell is doing the swirly and that Microsoft isn't, is
that at Novell the constant refrain is, Oh, yeah, we were almost right
about what would happen next quarter, except for this one little
uncontrollable factor that we didn't count on, but next quarter. . . " How
many quarters in a row does it take to at least have it dawn on you that,
if nothing else, you obviously aren't very good at predicting what's going
to happen to you in the next quarter?
> IF we sell x number of units THEN we will be profitable and healthy.
> From this rule we will focus on the number of units sold, and take that
> to
> be a key indicator of how healthy our business is. The prejudice of our
> theories often cuts us off from other, perhaps more rich, ways of
> looking
Like what? If my business makes a profit, it is healthy. Hard, measureable
numbers. If my business doesn't, I'm going to use whatever wisdom,
intuition and common sense I have to form a theory about that problem and
how to change it. Then, after I apply my theory, if I still don't make a
profit, I'm going to say I was wrong, and think harder. But I'm going to
keep the measure. It's not the measure that's the problem.
Another for instance, one closer to home:
A technical support organization that doesn't answer customer calls
quickly isn't cutting it. Customers say so all the time, and they mean it.
There's a hard measure that does address fundamental, real goals. But you
don't fix bad queue numbers by calculating how many calls you need
technicians to take and how quickly and then rewarding that or setting a
quota. You do that, you get a toothpaste effect, and your customers just
find another reason to be mad at you because your techs give them poor
quality, blow them off, evade finding answers, but at least do it before
the third ring.
You solve such a problem with intelligence and deep understanding of the
dynamics of your business and context, but you then judge the solution
based on the numbers changing.
> Our measurement systems, while important, create a type of corporate
> myopia. We need to put them in their proper perspective.
I think the proper perspective is, anybody who worries primarily about
numbers when creating policy and strategy is butt stupid. And anybody who
ignores or explains away numbers when judging the effects or failures of
their policies is blind. The one thing I can't remember, is, at Novell was
it the stupid leading the blind or vice versa?
I'd say all decisions should be made based on rational thought and the
context of your knowledge about how things work in principle. And all such
decisions should be proven by betting on a definite, measurable outcome.
The definition of being wrong is not getting the outcome. Period. What do
you think?
[...end of Ben's friend's msg. ...Rick]
-- Benjamin B. Compton bcompton@enol.comLearning-org -- An Internet Dialog on Learning Organizations For info: <rkarash@karash.com> -or- <http://world.std.com/~lo/>