Similarly, what if, despite all your sophisticated algorithms, there is a hole in your gas tank, and it’s leaking? Now, all of the elements that your algorithms are factoring into its equations –– your driving speed and style, the hill climbs, the boat you’re towing –– are beside the point: You are going to run out of gas sooner than the system is predicting.

 

Similarly, if you’re figuring the cost of a driving vacation, how are you going to account, weeks or months in advance, for recent wild, unpredictable fluctuations in gasoline cost –– from, say, $3 to $5 and back down under $2?

 

So, I maintain that models can never incorporate all of the significant variables that can influence an outcome. There is a “soft side” that optimization engines can never address –– like that hole in the gas tank, a buyer who just happens to dislike you personally, a retail company that simply doesn’t want to do business with you because your company cut cases or took a price increase.

 

Your model simply cannot know, or account for, any of these things.

 

In short, directional? Yes. Reliable? Sometimes. But absolute? No. There are simply too many variables that cannot be factored in by a computer model.

Finally, Dale, you say you are reducing complexity. In point of fact, all you’re really doing is moving it — and perhaps even increasing it in your overall system.

 

Think, for example, about a salesperson who uses these tools to optimize his trade spending. These tools may simplify his life –– but they are complicating life somewhere else: They’re increasing complexity for corporate staff, pushing it to analysts in your organization. To go back to your gas-gauge analogy, you moved the complexity from a Ph.D in your passenger seat to a little computer chip. That may be a cost improvement –– you don’t have to take the chip to lunch. But that chip has its own needs and limitations.

 

Remember, too, that if you push this complexity inside, you’re pushing it further away from the business. People sitting in an office in front of a computer terminal simply are not out there seeing what really happens, on the street in the real world –- all those messy customer relationships, competition, market forces –– and how and why these affect the business.

 

So again, the complexity itself –– and its cost –– has simply been shifted.

 

DALE HAGEMEYER: I’m okay with a computer chip, as opposed to the average salesperson, taking on complexity. As we all know, most account people would much rather be out managing relationships, selling products and programs, than trying to interpret data.

 

In my view, the right predictive models are a game changer.

Let’s look at what happens to create business. The average sales call hinges on three things:

 

n

“How much MORE do you have to spend with me this year?” (The customer’s assumption is that of course, you’re not going to keep your investment flat.)

 

n

“What new items do you have to introduce?” (The customer wants some NEWS out there.)

 

n

“What’s national support going to look like?”