How many metrics do you need? The metrics rule of minimums.

Ruler.jpg

The Metrics Rule of Minimums: The number of metrics needed is the minimum to accurately indicate state. There are a quite a large number of metrics that can be generated from a single project that could end up in a status report (I'll dig into details in a subsequent post). As the head of Onboard Informatics' Project Management Office (PMO), I am responsible for developing reporting and measurement standards for ALL projects. There are just too few hours in the day to maintain all those measurements and reports! I needed a guideline for our group to limit metrics development and maintenance effort but report status accurately. The metric rule of minimums seemed to fit the bill.

Metrics development is an organic process. A large pool of candidate metrics are developed and then from there you can whittle down the list significantly. Ideally, the final list should be fairly limited, only including the "right" numbers which depends on the activity at hand. This is followed by actual measurements in the field. The team can start early on identifying the metrics that correlate best with the state of the project. There is no reason to wait until the activity is fully over. If the team feels that metrics do not reflect the state of the work effort, then they can brainstorm other metrics based upon the perceived gap. Or they can revisit the original "large" metrics pool for other candiates. If the team feels an original metric is adding little value over time, then it can eventually be dropped.

No matter what type of work you have on your plate, you should identify the absolute fewest possible metrics to give managers and stakeholders an accurate state of the universe. Good, bad or ugly... metrics should reflect reality; otherwise, the effort to produce them is a waste of time. Worse still, if the "right" metric is missing, there is substantive risk that a project may come off its tracks without warning!

The "metrics rule of minimums" provides the following key benefits:

1. Rapid Communication of Status - due to limited number of data points. 2. Accurate Communication of Status - due to the "right" set of data points.

Frankly, if you have a history of trust between you and your manager, and your work is steady and predictable, then there may only be one simple Boolean metric required: "Is work on track - Yes or No". If only it were that simple...

By the way... this cannot be an original concept. That said, I distilled this little gem out of readings, web searches and some painful thinking. If anyone finds other references, please post a response. Thank you

Image Credit: Scott Akerman on Flickr.com