The following article originally appeared in The Harvard Education Letter (volume 23, number 3). Copyright 2007 President and Fellows of Harvard College. All rights reserved.

In our work on instructional improvement with low-performing schools, we are often asked, “How long does it take?” The next most frequently asked question is, “We’re stuck. What should we do next?” In our roles as facilitators of communities of practice focused on instructional improvement, in our work on internal accountability (Richard) and using data (Liz), and in our research, we have noticed some distinct patterns in the way schools develop as they become more successful at improving student learning and measured performance. Here are a few of our observations.

There are no “breakthroughs” or dramatic “turnarounds” in the improvement of low-performing schools. There are, however, predictable periods of significant improvement, followed by periods of relative stasis or decline, followed again by periods of improvement. This pattern of “punctuated equilibrium” is common across all types of human development: individual, organizational, economic, and sociopolitical.

A very low-performing school may see significant improvements in students’ scores in the early stages of concerted work to improve instruction. These early periods of growth are almost always the result of making more efficient use of existing resources—instructional time, teachers’ knowledge and skill, and leadership focus. For example, a school might extend time spent on math from 45 minutes a day to 60 minutes, or might make smaller groups for literacy instruction. Not surprisingly, the improvements in performance that occur as a result of improvements in existing resources are relatively short term. They are usually followed by a period of flattened performance.

If a school is on a significant improvement trajectory, this plateau usually represents a process of incorporating new knowledge into the previous base of knowledge and skill. The school that extended time spent on math might now focus on what the math instruction looks like—how to teach mathematics so that students have a conceptual understanding of the math rather than only a procedural understanding. These changes are, by their very nature, extremely challenging. They challenge teachers’ and administrators’ existing ideas about what it is possible to do. They raise difficult questions about the effectiveness of past practices. They require unprecedented investments of time and energy. And often they do not produce immediate payoffs in measured student performance.

In our experience, most of the learning that schools do occurs during the periods of flat performance, not during periods when performance is visibly improving. Periods of visible improvements in performance usually occur as a consequence of earlier investments in knowledge and skill.

Surviving the Slumps

Periods of flat performance in the improvement cycle raise some of the most difficult challenges educators face. It feels horrible when you and your colleagues are working harder than you have ever worked, when you have accepted the challenge of incorporating new practices into your work with students, when you are participating in planning and collegial activities that force you to move outside your comfort zone—and you see no visible payoff for these huge investments. These are the periods when it is important to develop a supportive work environment and positive leadership.

We’ve observed several practices in schools that thrive through stages of flat performance:

(1) They expect the flat periods and persist through them

(2) they have a theory about how what they’re doing will result in improved student performance

(3) they develop finer-grained measures for detecting improvement

(4) they make adjustments when evidence suggests that their efforts really aren’t working.

Expecting the flats and persisting. As schools gain experience with cycles of improvement and stasis (or decline), they recognize that the process of school improvement is the process of uncovering and solving progressively more difficult and challenging problems of student learning, which in turn demand new learning from adults. Once the initial gains have reached a plateau, teachers and administrators may begin to focus on a particular set of problems, often associated with broad categories of students, that require deliberate changes in practice.

For example, schools might determine that students are struggling with high-level thinking. One school might respond to this problem by focusing on the tasks teachers are asking students to do every day in the classroom: Are students being asked to do high-level thinking on a regular basis? What do high-level tasks look like in different subjects and grade levels? Another school might respond to the same problem by focusing on questioning: What kinds of questions are teachers asking in class? How might teachers incorporate more high-level questioning into their instruction? Another school might notice that teachers are framing high-level tasks and questions, but not checking to see whether students understand them. This school might focus on appropriate forms of in-class assessment. It takes time for these new practices to mature and become part of the working repertoire of teachers and administrators. Schools that are improving recognize and allow for this time and don’t switch gears if they don’t see immediate results on state tests.

Having a theory. It’s a lot easier to stay the course if the course is something you anticipated. As educators gain experience, they are more able to explain how what they’re doing will lead to the results they want and choose professional development approaches accordingly. We’ve seen this trajectory in schools’ use of the professional development strategy of coaching. At first, schools and districts may adopt coaching because it’s a popular strategy and they think that teachers need support around instruction, which coaches can provide. Coaching often doesn’t provide the hoped-for outcomes, however, until the school can articulate a theory about how the coaching is supposed to help. For example, if the theory is that coaching helps by modeling good instruction and that teachers who see this instruction will adopt that practice, which will then lead to student learning—all that is examinable. Does the teacher’s practice change after the modeling? Is there evidence of a difference in student learning? Having a theory also helps identify what improvements to look for in the gap between working hard and seeing state test results, so that you know whether to persist or change course. (For the record, our experience is that modeling alone rarely leads to change in instructional practice, but the point here is to have a theory that both shapes what form your action takes and is testable.)

Developing finer-grained measures for detecting improvement. In our experience, changes in student performance lag behind changes in the quality of instructional practice. Improvements are typically visible in classrooms before they show up on external measures. Improvement is not always as obvious as we would like, in part because we look in the wrong places (annual state tests rather than the daily work of teachers and students in classrooms); in part because we use tools that are designed to detect big changes, rather than the tiny ones that lead to the big ones (the equivalent of using a clock with no second hand to measure improvement in the speed at which you can run a mile); and in part because sometimes things get a little worse before they get better. We see this last pattern frequently when teachers go from asking students questions to which there is a correct answer to asking questions for which there are multiple possible answers. At first, teachers aren’t very good at asking the questions or setting up a classroom environment in which ambiguity and intellectual risk-taking are valued, and students aren’t very good at providing answers that require sentences rather than two-word responses, or at offering rationales for their answers.

Visible measures of progress are critical for motivating and encouraging educators to persist in the challenging work of improvement. Even the most dedicated and optimistic among us will stop if there’s no sign that what we’re doing is making a difference, or might make a difference eventually.

Making adjustments. In fact, schools that are improving do stop if there’s no sign that what they’re doing is making a difference. Having a theory and the right tools to test it makes it possible to identify the need for adjustments. Improving schools are willing to make adjustments, including stopping a course of action, if over time the evidence suggests their strategy isn’t working.

The Next Level of Work

Sometimes, however, schools aren’t sure what adjustments to make. What should schools do when they get stuck? “Stuckness” typically happens when people feel like they are doing their best work and it’s not paying off in visible evidence of improved student performance. Billie Jean King—perennial tennis champion and accomplished coach—describes the transformation that occurred in her own career when she learned to regard errors as “feedback.” This turned her frustration into reflection, and her reflection into increased focus and correction. Evidence that our best efforts are not producing what we want them to produce is feedback. The evidence is trying to tell us something about what we are doing, and if we listen to it, reflect on it, and give it voice, it will help us understand what to do next.

In our work, we help practitioners frame the next level of work by examining what they are currently doing, looking at evidence of student learning for clues about what is strongest in their practice and where they might see opportunities for improvement, strengthening the capacity of colleagues to work collectively on instructional issues, and increasing the specificity, or “grain size,” of the instructional practices they are working on.

It is not unusual for schools to be doing very good work in a given content area—math or literacy—and for that work to be manifested in visible improvements in student performance. As time passes, however, teachers and administrators discover that what they considered to be their “best” work is not reaching certain students, or that performance overall is stuck in the middle range and not moving into the advanced range. These kinds of problems typically require closer examination of what students who are doing their “best” work are actually doing. What teachers typically discover is that the actual tasks that students are being asked to do, while considerably more challenging than those they were previously asked to do, are not at a level that will lead to the kind of student performance that teachers hope for. Or they find that the tasks are challenging, but the work is not scaffolded in a way that allows students to reach higher levels of performance. Or that students in some classrooms are able to do challenging tasks, but comparable students in other classrooms are not. The next level of work in each of these situations is different.

Improvement and Accountability

As schools improve, three different but related processes are occurring. First, the level of knowledge and skill that teachers and administrators bring to the work of instructional practice is increasing. Second, teaching is moving from an individual to a collective activity, and internal accountability—the level of agreement and alignment across classrooms around powerful practices—is increasing. Finally, the school is aligning its organizational resources around support for instructional improvement.

All of these processes take time. And, as noted above, they do not occur in a straightforward, linear way. Just as with individual students, individual schools really do differ in the challenges they face and in their capacity to incorporate new practices.

Our accountability systems, as they are currently designed and implemented, do not reflect the real demands of school improvement. Well-designed accountability systems would start from an empirical knowledge of what school improvement looks like when it’s happening and establish incentives and supports that accord with that knowledge. At the moment, the process is reversed: Accountability systems establish arbitrary timetables and impose powerful negative incentives on school improvement without any grounding in knowledge of how the process occurs. People in schools are forced to invent the knowledge themselves and must often work against the structures and incentives of the accountability system in order to get the job done.

The discipline of school improvement lies in developing strong internal processes for self-monitoring and reflection—not in meeting an artificially imposed schedule of improvement. That existing accountability systems don’t reflect this reality is one of the great political tragedies of current education policy.

So, how long does it take? Educators know deep down that this is not the right question because it implies a finish line or a summit that we will someday reach. But that’s not how improvement works. Some days we may feel like Sisyphus, forever pushing the boulder up the mountain, never to reach the top. But other days we get to what we thought was the summit and realize that still greater things are possible, things we couldn’t see from below.

This is why we teach and lead. Improvement, after all, is essentially learning.