This message is available on the new blog from this link
is in the most total humiliation that our war is the average of the distributed battlefield during the last fight. .. It became aware of its weaknesses: it is aided by his two companions, Minimum and Maximum, she decided to return to the front ... ;-) And yes ...
During my last demonstration I set the default using the average to report a distribution (a set of data). But the average has two indicators that could help account for a data set: the minimum and maximum. So can we use the average if it is accompanied by the minimum and maximum indicators ...!?
To try to answer them I will rely on the memory of an error management system that strongly influenced me.
The head of a purchasing department had to define the goals of his team for the upcoming year. The objective of the previous year was based on the cumulative service rate to the year (the service rate shows the fraction of orders that are delivered). For our example we assume the service rate target last year was set at 97% (ie that on average: 100 lines of when orders are placed by customers, 97 will be delivered, the other 3 undelivered lines corresponding to stockouts).
Here is the result of the year just past ...
As you can see, there were large variations during the year. And our manager remembers well the pressure he had suffered on the part of clients and its hierarchy, during weeks 20 to 24 ... He did not want to relive it, so he decided to set two goals for his team: the first will remain on the annual average (97% of its proud, he stood on it), and the second would cover a maximum rate of only minimum service to not exceed (say 96 %)... I saw two constraints in the choice of this indicator:
- First: when the threshold is exceeded: the teams are unmotivated because the goal would be missed and premium objective with. :-(
- Secondly: weeks with a very low level indicator make it more "vulnerable" to the inevitable failures (the delivery problem supplier) and inversely surges of activity could cause disruptions and technical lead exceeding the threshold.
In summary the calculation was not representative of the variation in the rate of service or motivating the team to Purchaser.
So I proposed a target based on the standard deviation and increased service levels as shown in this graph:
The objective was to reduce the standard deviation (corresponding to the average change) half (the orange line past the previous graph line on the light green graph). And reassess target at 97.5% for average annual service rate (dark green). Knowing that after a short study, the removal of outliers that were the weeks 20 to 24, easily could achieve this new rate. Thus reducing the average change was a direct result of an improvement in the rate of annual service.
I saw two well done:
- The measurement of the change on the whole of the motivation remains.
- Improving service levels to improved overall customer satisfaction, hierarchy and department employees who could be proud of having achieved a real goal.
What the manager decided:
Following my suggestion, the manager was trying to validate this type of calculation in setting its target ... only problem: he did not master the math ...!
And considering that he was unable to master the calculation of the indicator he shortens it wrong he was unable to control the indicator and thus the aim of its staff ... So he set his minimum goal by adding a tolerance threshold is exceeded over a week.
Epilogue: In mid-year heat wave came ... Product sales Mosquito exploded ... Suppliers were unable to meet market demand ... The suppliers were very well done Leut work ... and had no premium on their goal ... but the main thing was this: they could explain it!
This time: the Min and Max have won the battle because they blew the market away type ... but admit that the case is rather unfair! is not it?!
Will it one final battle to decide the ...!? ;-)
0 comments:
Post a Comment