I remember the days when I was in charge of predicting arrival times at milestones for a large heavy construction project.
Each week, my group would collect progress data on some 50 engineering/construction contracts and my job was to run it through the computer and generate reports that would show actual progress and project early/late arrival at project milestones.
The usual outcome was a contract would start off with reasonable contingency, only to start showing problems early on in the execution. Management would complain, the status of the project would immediately “improve”, only to have projections revert back to the old ones.
The frustrating thing was the members of our Scheduling department could graphically extend trends before “corrections” were made and fairly accurately predict eventual outcomes. One of the metrics we developed was if a project got into trouble during the 1st twenty percent of its timeline, the chance of on-time completion would be very low indeed.
I see the same type of behavior at Balanced Scorecards – interference with system algorithms results in management seeing what they want to see and performers getting maxed-out incentive payments.
While we are on the topic of Balanced Scorecards, it pays to look behind the façade.
There are three stages to any activity, the past, the present and the future.
We cannot do much about the past except to learn what to do/what not to do on future projects. As for the present, it is what it is, providing the information being looked at is complete and timely and has not been tampered with. And, regarding the future, well, all those who have the ability to predict the future have long ago retired to private islands and are not available to help us with predictions.
Scorecards have some value, but staring at a fixed set of metrics that have a foundation on manipulated data is not the best way to run a corporation.
The approach I recommend is develop your strategies/objectives/metrics in a graphic multi-root tree environment (somewhat like an extension of traditional one-hub-multi- spoke mind maps), but consolidate transaction level run-time operations data up to this environment so that it becomes the corporation’s KnowledgeBase.
People will go to the Kbase for knowledge, they will develop strategies/goals in the Kbase, they will be able to set time slices and view “status” along a continuum that stretches from the past, to the present to the future (tempered by risk and uncertainty assessments).
But, go easy on incentive payments unless you have an army of independent auditors validating algorithms/ consolidations.