Leading Education Systems With Purpose in an Incoherent Environment

You can turn data into good information that supports action

Education leaders are constantly being asked to make coherent decisions in an incoherent environment.

That may sound dramatic, but it describes the daily reality of many who lead large school systems. They are trying to make sense of artificial intelligence, technology and screen-time debates, assessment proliferation, through-year assessment proposals, accountability results, budget constraints, staffing shortages, and public pressure for improvement. Each topic often arrives as its own urgent problem. Each comes with its own vocabulary, vendors, advocacy groups, policy pressures, and implementation questions.

These are stress tests of system coherence and not separate fires.

I was reflecting on this recently as I prepared a keynote for district leaders, but the problem is not limited to districts. State and district leaders both function in incoherent environments and are both responsible for turning evidence into useful information for action. To move from information to action, making coherent decisions—that is, making choices that logically fit together—is important.

Leaders are often asked whether they should adopt a tool, restrict a practice, redesign an assessment, or change a reporting system. But the better questions are: Can the system respond with purpose, sound interpretation, and coherence? Can leaders clarify the decisions they are trying to support? Can they distinguish a useful signal from a more complete diagnosis? Can they help educators, families, communities, and policymakers understand what the evidence does and does not say? And can the system support action after the signal is visible? Below, I offer three example issues that demonstrate a pattern. 

Issue 1: Start With Purpose, not Pressure

When a new issue hits the system, the temptation is to move quickly to a solution: an AI policy, screen-time limits, an accountability response plan, or a new testing model.

Sometimes those responses are necessary. But speed can hide a basic design problem: We have not always clarified what decision those responses are supposed to support.

Different decisions require different evidence. A tool that monitors broad system trends is not the same as one that diagnoses student misconceptions. State summative assessments can provide a common, technically defensible signal for monitoring, transparency, evaluation, and public engagement, but they are not daily instructional tools. As several of my colleagues have written, the question is not whether test data are useful in the abstract; it is whether they are useful for specific decisions.

This is why balanced assessment systems matter. Balance is not created by adding one state test, a few interim assessments, and some classroom checks. As I argued in “A Recipe for Balanced Assessment Systems,” assessments do not create balance; use creates balance. Leaders also need to identify what evidence is missing, which interpretations are warranted, and where strategic abandonment may be necessary.

Issue 2: Protect Interpretation

Data do not become information on their own. People turn data into information through interpretation.

That is where systems often succeed or fail. A score, rating, dashboard, color code, performance level, or growth result can focus attention. But the interpretation of that signal determines whether the system learns from it, overclaims from it, or ignores it. Many of the most consequential mistakes in assessment and accountability do not come from the mere existence of data. They come from claims that outrun the design.

Consider through-year assessment. It remains attractive because different versions of it aim to address long-standing challenges with state summative testing: long tests at the end of the year, results that arrive late, and data that can feel distant from instruction. These are real problems, but solving a timing problem does not automatically solve an interpretation problem. 

If a through-year system is expected to support both accountability and instruction, leaders need evidence that the resulting scores can support those different uses. As Through-Year Assessment: Ten Key Considerations emphasizes, these models involve tradeoffs across design, curriculum, instruction, logistics, technical quality, and accountability use. Solving a timing problem does not automatically solve an interpretation problem.

The same caution applies to technology and screen-time debates. More technology is not automatically better; less technology is not automatically better. Recent work from Meredith Coffey and Erin Mote points toward harder questions about purpose, quality, safety, evidence, and support—not just how much technology students use.

Issue 3: Accountability Is a Signal, not the Strategy

Accountability can focus attention. It can help identify where performance is strong, where support may be needed, and where patterns require public explanation. But accountability does not improve educational systems by itself.

A stronger signal is not the same as a stronger improvement system.

When accountability results are released, people usually ask familiar questions: What does this mean? Is this fair? Who is responsible? What do we do now? How do we explain it? These are broader system-design questions that go beyond communication. 

If we expect accountability to support improvement, the system must be ready for what happens after the signal: interpretation, diagnosis, role clarity, support, action, and monitoring. In Reclaiming Accountability, Laura Pinsonneault and I argue for moving from measurement and labeling toward learning and improvement. That shift requires a clearer theory of action: What is supposed to change, who is supposed to act differently, what support will they need, and what evidence would show whether the change is happening? Without that clarity, accountability can become a signal without a strategy.

The Pattern: The Risk of More

Across these examples, there is a pattern: Systems often respond to complexity by adding more data, dashboards, reports, assessments, prediction, meetings, guidance, and tools.

Some additions are necessary. But more does not automatically mean better. Faster does not mean clearer. More detailed does not mean more useful. More visual does not mean more meaningful. More predictive does not mean valid for every use. And more public does not always mean more transparent.

Coherence is not necessarily created by adding something. Sometimes it is created by subtracting, simplifying, or redesigning routines so people can make better use of the information they already have.

The Action: Three Responsibilities for Education Leadership

State and district leaders are not just consumers of data from the state. They influence what is measured, how it is reported, how it is interpreted, and what supports follow.

That influence creates three responsibilities.

1. Protect purpose. Before adopting a new tool, report, assessment, dashboard, policy, or meeting routine, ask: What decision is this information supposed to support? In practice, require any new addition to name three things: who will use the information, what decision it is meant to support, and when that decision needs to be made.
2. Protect interpretation. Before acting on results, ask: What does this evidence show, what does it not show, and what else do we need to understand before assigning causes or choosing a response? In practice, pair major reports or result releases with a short “claims and limits” statement: What we can say, what we should not say, and what additional evidence we need.
3. Protect coherence. Before adding another layer to the system, ask: What should we stop doing, simplify, combine, or redesign so people can use the information they already have? In practice, pair every major new request with a reduction decision: what will we stop, shorten, combine, automate, or retire to make room for this work?

These routines are not meant to slow systems down for their own sake, but rather to make action more intentional and better supported by relevant evidence. In large systems, clarity cannot be treated like a luxury; it is a condition for trust, usefulness, and sustained improvement.

The final test of an information system is whether the right people can use the information well.

So a key question for state and district leaders is this: What better question are you helping educators ask and act on?