Perspective on the PerformanceStat Potential: A Leadership Strategy for Producing Results

By Michael J. Keegan



It started two decades ago with CompStat in the New York City Police Department and quickly jumped to police agencies across the world as well as to other agencies within New York City. It was adapted by the City of Baltimore, which created CitiStat, the first application of this leadership strategy to an entire jurisdiction.

Today governments at all levels employ PerformanceStat, a focused effort by government executives to exploit the power, purpose and motivation, responsibility and discretion, data and meetings, analysis and learning, feedback and follow-up, all to improve government’s performance.

  • What is PerformanceStat?
  • How is PerformanceStat a leadership strategy for producing results?
  • What are the core leadership principles and key operational components of this leadership strategy?

Professor Bob Behn of the Kennedy School at Harvard University, and author of The PerformanceStat Potential: A Leadership Strategy for Producing Results, joined me on The Business of Government Hour to share his perspective on these questions and to discuss his definitive book about the PerformanceStat phenomenon. His research underscores that, for the PerformanceStat leadership strategy to work, it must be tailored to meet an organization’s specific circumstances. To realize the PerformanceStat potential, according to Behn, government executives have to make an explicit—and thus very dangerous— commitment. It is about taking the responsibility to achieve specific public purposes but, for government executives, embracing this responsibility puts them out there with no place to hide, as Behn explains in the perspective he shares. The following is an edited excerpt of our conversation.

Sixteen Causal Behaviors Underlying The PerformanceStat Leadership Strategy

  • Behavior 1: Reiterating the purpose can keep everyone focused on the big picture.
  • Behavior 2: Analyzing data can reveal significant performance deficits.
  • Behavior 3: Creating targets can specify exactly what needs to be accomplished by when.
  • Behavior 4: Making operational assignments can define who needs to fix which performance deficits next.
  • Behavior 5: Devoting resources and time to PerformanceStat can dramatize the chief executive’s personal commitment to improving performance.
  • Behavior 6: Conducting meetings can focus everyone’s attention on what is most important.
  • Behavior 7: Requesting reports on progress can ensure that targets and assignments are taken seriously.
  • Behavior 8: Asking questions of individual subunit managers can promote personal responsibility.
  • Behavior 9: Following up frequently on targets and assignments can create the feedback that can suggest adjustments.
  • Behavior 10: Distributing comparative data widely can help every team appraise, without delusions, its own performance.
  • Behavior 11: Scrutinizing the positive deviants can facilitate everyone’s learning.
  • Behavior 12: Recognizing accomplishments publicly can confirm that success is possible and valued.
  • Behavior 13: Reproving the recalcitrant can get everyone’s attention.
  • Behavior 14: Telling stories can foster a results-focused culture.
  • Behavior 15: Abetting everyone’s implicit evaluation of everyone else can breed individual and team motivation.
  • Behavior 16: Remaining persistent can prove that this isn’t going away.

Source: Bob Behn’s Performance Leadership Report

PerformanceStat as a Leadership Strategy

PerformanceStat is a very complicated concept, so perhaps it would help to highlight some of its specific components. First of all, I want to emphasize that it’s not a system and it’s not a model or leadership strategy. It’s not solely a performance measurement either. In fact, I use the phrase “performance leadership.” From my perspective, too many people use the terms performance management and performance measurement interchangeably, as though, if you do the measurement, the management stuff automatically happens.

PerformanceStat is a leadership strategy that is designed to fulfill specific public purposes by producing specific results; that is, there are some defined results trying to achieved based on a government executive having a specific purpose in mind that dovetails with their agency’s mission.

One may ask: what can be done to make this happen? One practice is to have a series of regular integrated meetings to discuss what’s effective, identify specific problems, the causes of the problems all the way developing strategies that might help us solve a particular problem. In these meetings, follow up on previous targets and commitments are reviewed and progress assessed. During these meetings, we always try to figure out what the next issue is that we can address and how to bring the PerformanceStat approach to bear as a leadership strategy.

CompStat and Its PerformanceStat Progeny

Manifestation of this leadership strategy starts in New York City Police Department in January of 1994. Rudy Giuliani is elected mayor in November of 1993 after campaigning on the key platform of reducing crime in New York City. He hires Bill Bratton to be his police commissioner. Bratton comes in, hooks up with his old buddy Jack Maple and they create this thing called CompStat, which focuses attention on specific categories of crime and then they dedicate resources to reducing crime in particular areas. They have a series of meetings with the city’s precinct and the bureau commanders, during which they work to identify problems and try to identify strategies that will reduce crime of particular types in particular areas.

Other police departments copied the CompStat strategy to reduce crime while other agencies in New York City that had nothing to do with crime or policing also began to adopt this focused approach. For example, the NYC Human Resources Administration created JobStat, which tracks job placement for welfare recipients. The NYC Administration for Children’s Services also created ChildStat, which focuses on reducing the problem of child abuse. As a number of New York City agencies employed this approach, its adoption spread across the country.

The city of Baltimore took the PerformanceStat approach to the jurisdiction-wide level applying this strategy to every agency in the form of CitiStat. The last iteration is at the federal level in which, even before the GPRA Modernization Act began requiring agencies to run quarterly performance reviews, the U.S. Department of Housing and Urban Development established HUDStat, the Federal Emergency Management Agency created FEMASTAT, and the Food and Drug Administration created FDA-Track; all of which use the same basic strategy.

In the end, though, government executives need to figure out how to adapt the PerformanceStat potential to their specific and unique circumstances. You can’t simply take what is being done for the police department or a social services organization and plop that format into any organization. You have to adapt the basic concept and principles to the new circumstances, which require you to understand what those underlying principles are. I’d argue that those principles are really leadership behaviors. They are actions leaders take to motivate and drive people to accomplish specific public purposes by achieving specific results. Often, as a motivational tool, we give people a target to hit. If people have a target to hit, they’re much more likely to focus their attention on it.

PerformanceStat is not a System or Model

All too often, PerformanceStat is referred to as a system, perhaps a model, or a tool. These are misplaced metaphors because they present a very mechanistic way of running an organization. It suggests that one can set up such a system in an organization, push the start button, and walk way. Unfortunately, the overuse of the systems metaphor, many efforts to implement seem to miss something critically essential. PerformanceStat is not a system, or a model. It is a leadership strategy. For to achieve the strategy’s potential to produce real results requires active leadership. Moreover, the leadership team must adapt the strategy to fit its specific public purposes.

PerformanceStat is more than a system or model, more than data and accountability. It is a leadership strategy, and in any organization, improving performance requires real leadership.

PerformanceStat’s Five Visible Features

  1. The room in which the PerformanceStat meetings are held.
  2. The data that form the basis of the discussion at the meetings.
  3. The technology used to analyze and project these data.
  4. The staff who manage and analyze the data.
  5. The meetings themselves.

Defining PerformanceStat

I have crafted a definition of PerformanceStat: A jurisdiction or agency is employing a PerformanceStat leadership strategy if, in an effort to achieve specific public purposes, it holds an ongoing series of regular, frequent, integrated meetings. During these meetings, the chief executive and/or the principal members of the chief executive’s leadership team plus the director (and the top managers) of different subunits use current data to analyze specific, previously defined aspects of each unit’s past performance. They conduct this analysis to follow up on previous decisions and commitments to produce results. Thereby, they examine and learn from each unit’s efforts to improve performance, solve performance-deficit problems, and set and prepare to achieve the next performance targets.

Performance Deficits

A performance deficit, at least as I use the phrase, is something along the value chain, from inputs to activities and processes, to outputs to outcomes that isn’t quite working right and that we need to fix. The problem could occur very early in the value chain. If we fix this problem, things will get better. In fact, in many organizations, what we do early in the value chain is more critical than what we do later because it affects everything else. We have to think about what performance deficit we need to fix. I will give you a simple example, one that doesn’t involve PerformanceStat, but illustrates it and is clearly an example that everybody can understand.

The focus is on Lowell, Massachusetts, an old mill town in which the mills have long since been shut down. The old buildings are now occupied by immigrants. I have probably told you enough to allow you to figure out what the test scores for the children in the elementary and secondary schools are in this city. Lowell decided that one of its key performance deficits was its truancy problem. Too many kids weren’t in school so the city leaders organized, not just the schools and police, but community organizations and parents, to get the kids in school. Some people had the responsibility for calling up well-known truants at six-o’clock in the morning to get them out of bed and get them to school. So, eventually, they get kids into school and as time goes on, test scores start going up. Now, the theory behind this isn’t too complicated. If the kids aren’t in school, they aren’t going to learn or to the extent that they do, it’s not the types of lessons we want them to learn.

This is an example of a performance deficit that needs to be fixed. People start to concentrate on this issue, set targets for achievement, hold meetings, identify what progress has been made and what else needs to be done, and then address all of these problems by reporting back at the frequent, integrated meetings we’re holding.

The Evolving Nature of PerformanceStat

PerformanceStat 1.0 is about creating operational effectiveness. It requires managers and employees—with prodding from the organization’s leadership team—to do the basic assigned tasks in a timely and proper way. It is about production: getting the core outputs done and on time.

PerformanceStat 2.0 is about redesigning the existing core functions. It requires managers and employees—with prodding and guidance from the organization’s leadership team— to rethink how they go about performing their core functions. It is about identifying new, innovative ways to produce the existing, well-established outputs.

PerformanceStat 3.0 focuses on outcomes. It requires the managers and employees of multiple agencies—with the prodding, guidance, and active analytical engagement of the leadership team—to figure out who needs to collaborate with whom to produce the desired outcomes. It is about assigning priorities to outcomes and then motivating everyone to figure out how best to achieve them. When governmental jurisdictions and public agencies graduate from PerformanceStat 1.0, to 2.0, and then to 3.0, they move from ProductionStat to OutputStat to CollaborationStat. But before they can make such collaboration work, they have to learn to fix the potholes.

The next story I tell is about my son and me standing at the bottom of Tuckerman’s Ravine in New Hampshire in the White Mountains. He was pretty young. We’re going up to the League of the Clouds Hut and he looks up and says “I’m not going to make that.” I look up and think to myself, “Wow, it’s a long way!” but I don’t tell him that. We made it to the top but we didn’t jump. We took the challenges one step at a time, and that’s what you have to do here. You can’t say, “Oh, I could just get there tomorrow,” because you can’t go it alone; you have to bring your organization along. The people in your organization have to believe that they can get it done. They have to recognize when they’ve been successful and say, okay, now that we’ve been successful, maybe we can be successful again at the next level.

Conveying Tacit Knowledge

With my book, The PerformanceStat Potential: A Leadership Strategy for Producing Results, I try to convey tacit knowledge in an explicit way. Explicit knowledge can be conveyed in words, blueprints, and forms that people can follow explicitly, but tacit knowledge cannot. Teaching someone how to ride a bike illustrates my point. The physics of bicycling can be conferred through explicit knowledge, while riding one requires tacit knowledge. Explicit knowledge is not required to ride a bicycle—or even to create one. Knowledge about exercising leadership is similar to knowledge about riding a bicycle. It is not explicit, but strictly tacit. You can no more teach people to be leaders than you can teach someone to ride a bike. You can only help them to learn.

An example I use in the book is how successfully Scotland employed the CitiStat strategy. The government sent people to Baltimore to see it firsthand. I was struck by how successful Scotland was in implementing CitiStat I asked how the government did it. Well, its leaders didn’t send somebody to watch one meeting for an hour. They sent a team to investigate the Baltimore CitiStat process for a week. The Scots talked to the agencies and staff and watched many meetings. They sent observers to Summerville, Massachusetts to find out how that jurisdiction implemented it. They absorbed themselves in the activity, so they could figure out what the tacit knowledge was and how they would make it work in an obviously very different set of circumstances. I would encourage anybody who is going to use this strategy to see it in action. Visit and closely observe the use of PerformanceStat in multiple jurisdictions over a decent period of time.

Unfortunately, unless you are willing to spend a week poking around, observers of any example of PerformanceStat in action are unlikely to be exposed to the tacit knowledge on how the behavior of the leadership team affects the performance of staff. The reliance of an effective PerformanceStat on the tacit knowledge of the members of the leadership teams explains why I call it a “leadership strategy” and why it’s incorrect to view PerformanceStat as a mechanistic system.

Key Organizational Competencies

Basically, an organization looking to realize the PerformanceStat potential needs some foundational competencies. The first competency is a results-focused mindset. We are focused on actually accomplishing specific purposes. The second competency is an analytical bent. Some people are very analytical. They can look at data or other types of information and try to learn from them. It’s that analytical learning capacity that any organization needs to build. The next organizational competence is having a purpose-driven persistence. Are we going to be persistent? Are we going to focus on a purpose? It’s not just persistence but persistence toward achieving a particular purpose; persistence toward achieving a particular result.

Distinguishing a Real Strategy from a Pretender

I would say that, of all the things that separate a real strategy from a pretender strategy is, in a pretender strategy, the leadership doesn’t really care and it shows. Leadership is not engaged. Leaders say, “Hey you guys, go do that and we’ll go pay attention to other things.” If the leadership is really engaged, if they’ve made a commitment, they say “We’re going to produce these results and we’re going to get them done by this date.” Then everybody understands that they take this commitment seriously. Moreover, if they’re spending their time on it, if they’re actually investing their own most valuable resource, which is their time, then people will begin to take it seriously.

If you and your leadership team are showing up at the biweekly or monthly or quarterly performance reviews, people say, “Oh, that’s interesting. The leadership actually cares. Maybe I should care too.” If they don’t show up, pretty soon everybody is sending their top interns to the meeting, which is characteristic of a pretender strategy.

Leadership Behaviors Underlying the PerformanceStat Potential

The first four leadership behaviors underlying the PerformanceStat potential involve focusing staff on what needs to be accomplished. The remaining behaviors are more about motivating folks to getting it done. I’ll highlight the initial four leadership behaviors. The first one is simply repeating and repeating what we’re trying to accomplish. You do this to make the objective clear and to make sure that everybody in the organization understands what our job is. The problem is, the first time you state the targeted accomplishment, people say, “Oh that’s really nice,” and they forget about it. The second time it’s repeated, people think, “Oh, they said that again. Oh that’s interesting.” You have to make the point really clear and you have to repeat and repeat and repeat it because people don’t take you seriously until you repeat it.

The next leadership behavior is analyzing the data to determine which performance deficits you have to fix next.

The third behavior involves creating targets; doing this tells people “this is what we’re looking to accomplish by this deadline.”

The fourth leadership behavior is making assignments. Don’t just have a target; give people specific assignments to do specific things to ensure that the target will be achieved.

A Leadership Strategy for Producing Results

It’s hard to make a one-to-one comparison on whether PerformanceStat actually improves results. You can’t run a gold standard double blinded controlled experiment, but what I’ve try to do in this book is to look at things that we know from behavioral literature, both individual and organizational behavior, to understand how people behave under certain circumstances within specific context.

What is the cause-and-effect theory behind PerformanceStat? As I’ve noted before, this leadership strategy focuses a very bright light on the performance of individual units, thus motivating each unit to improve and generating experiments from which everyone can learn.

In the end, however, no organization can improve its performance if it has no responsibility for its performance. No organization can influence results if it lacks the means for producing those results. No organization will make a specific improvement in performance until its leadership team commitments itself to specific improvement. It is key to understand that to realize the PerformanceStat potential, public sector executives have to make an explicit commitment to it.

Download this article

Download the entire Winter 2015 magazine