On why all public executives need to remember that picking measures is leadership job: analysts should not be asked to decide what better performance means. Picking performance measures is a leadership responsibility.

• Author: Behn, Robert D.
• Position: Commentary

The phone call was not unusual. I'd received such calls before. I'll receive them again.

The callers were from the analytical shop of a large public agency. The agency director needed some performance measures and these analysts had been charged with producing them. Thus, they were seeking some advice, a few suggestions, a little help thinking through what measures to recommend: What characteristics should their measures have? How many should they offer?

Isn't this weird? I had little more than a newspaper reader's knowledge of the agency. What kind of useful guidance could I offer?

The agency, of course, had data. Lots of data. It had data on internal processes. It had data from the organizations to which it distributed funds through grants and contracts.

Indeed, the analysts described their agency as "data rich." The agency had created almost a thousand measures that subunit managers could use. Yet, the agency collected data for only a few of these measures. It used even fewer.

Indeed, although the agency had created performance measures, it had not created an integrated effort to use the data. No one used the data to set policy. No one used the data to drive performance. As far as I could tell, no one (save for a few wonks) even looked at the data.

(I confess: I'm a data wonk. In fact, I am eagerly awaiting the 2011 baseball season's first box score.)

Rather, the principal purpose of collecting all of these data was the need to keep an overhead agency happy. Today, every governmental jurisdiction--as required by the background radiation now arriving at earth from the big bang--has an overhead agency that demands all of the line agencies produce performance measures. (Do such overhead agency's have their own performance measures?)

Naturally, the line agencies respond by creating analytical units with the specific task of feeding data to the overhead agency. (You, of course, have seen this behavior in action.)

But using these data to actually measure performance--to create an understanding of how well the agency (plus those organizations with which it collaborates through contracts and grants) is doing--is not straightforward. Even more challenging is the task of actually using such data to motivate and produce better results.

A prerequisite for improving performance is a determination of what better performance would look like. Someone (or some group) needs to decide what kind of improvement the agency needs to make next and how it will...