Consider this, a light bulb that has an expected life of 2000 hours, plus or minus 200 hours costs more than a light bulb that has an expected life of 3000 hours, plus or minus 1500 hours! That is because the first bulb is MORE RELIABLE. Reliable means that it PREDICTABLE. Perhaps many of Datek's customer problems come from lack of predictability. Some orders are executed in 3 seconds, which I think is reasonable, and other orders are not executed for a long long time. Yet, to the customer, there may be no reason to believe that one order is any different from another. Control requires the ability to predict and control is what we want.
Yes, predictions are difficult. However, that's why we get the big bucks.
It is not impossible, keeping in mind it does not have to be perfect.
The key is reducing the variance of the estimate. To do that, we must identify the independent variables. Please list the factors that come to your mind that effect the time required to fill an order. Then, for each variable, identify meaningful levels of each variable. For example, "Size of order." Now you should enumerate the levels of this variable (e.g., 1-99, 100-1000, 1000-5000, 5000+, whatever you think the meaningful breaks are) that will differentially influence the fill time. Do this for each variable. It may be difficult at first, but take a stab at it. I'll help you along the way.
|
