|
What's the best way to set targets for measures where you don't have past performance indicators (i.e., brand new measures). Does it make sense to just guess? Or will this lead to people trying to game the system? |
|
Kaplan has a way to set targets that he describes in the "Execution Premium" book. Basically, it's a percent achievement of "perfection" target. So, if you decide that "100% safety" is best in class, but you're not there yet, you can get some percentage of that goal each year, acknowledging that it takes some time to get there. |
|
I was just in a meeting at the American Council for Technology, Industry Advisory Council where we discussed this exact topic. One person suggested scenario planning where you try to figure out the best and worst outcomes and then setting your target based on the probability of the outcome happening. I think that can work where you have a good handle on the data and its behavior. My experience also shows that you might want to just collect data for one or two periods and then set your target based on your past results. Thus you will be in discovery phase for a little bit of time. |
