eduKateSG Learning Node Series · 0078

How Instructional Rounds Work | See the Pattern Across Classrooms Before You Try to Fix the System

A school can contain twenty excellent teachers and still have a system problem.

One classroom may have strong questioning, another strong modelling, another unusually good student discussion. Yet across the school, the same deeper pattern may repeat: students rarely explain reasoning, tasks stay at recall level, feedback arrives too late, or teachers are carrying most of the cognitive work while learners wait for the next instruction.

If leaders visit one classroom and immediately prescribe a solution, they may mistake a local event for a system signal. Instructional rounds are designed to slow that reaction down. A network of educators defines a problem of practice, observes the instructional core across multiple classrooms, records descriptive evidence, looks for recurring patterns and then identifies the next level of work.

Rounds are not about finding the best teacher or the weakest teacher. They are about learning what the organisation is repeatedly producing.

The 50-Second Read

• Instructional rounds are a collaborative observation-and-analysis process for school or system improvement.
• The model was developed in education with an explicit analogy to medical rounds: practitioners learn together by examining real work in context.
• A rounds cycle begins with a high-leverage problem of practice that the school can influence.
• Teams observe across several classrooms and focus on the instructional core: students, teachers, content and the tasks that connect them.
• Observers collect low-inference descriptive evidence before making interpretations.
• The debrief moves from description to patterns, then to predictions: if this pattern continues, what are students likely to know and be able to do?
• The cycle ends by naming a next level of work, not by producing a league table of teachers.
• Rounds are developmental only when they are clearly separated from hidden evaluation and surveillance.
• The research base is smaller than the popularity of the model may suggest; evidence includes promising implementation studies, but rounds should be treated as disciplined inquiry rather than a guaranteed intervention.

Canonical Owner Boundary

This page owns instructional rounds: the network-level cycle of defining a problem of practice, visiting multiple classrooms, describing what is observed, identifying system patterns and choosing the next level of instructional work. Teacher Professional Learning remains the umbrella owner for career-long educator development. Instructional Coaching owns focused improvement of an individual teacher’s practice. Lesson Study owns collaborative investigation of one research lesson. Instructional rounds ask a different question: what recurring instructional pattern is this school or network producing, and what collective work should happen next?

1. The Medical Analogy Is Useful—If We Use It Carefully

The term rounds deliberately evokes medicine. Clinical teams learn by examining real cases, comparing evidence and making professional reasoning visible. Harvard educators Elizabeth City, Richard Elmore, Sarah Fiarman and Lee Teitel adapted that broad logic to education in their work on instructional rounds.

The analogy has limits. A classroom is not a patient and teaching is not diagnosis in a biomedical sense. But one principle transfers well: complex professional work becomes easier to improve when practitioners can see it together, use a common language and reason from evidence rather than anecdote.

2. The Unit of Analysis Is Larger Than One Teacher

A traditional observation often asks, “How well did this teacher perform?” Instructional rounds ask, “What do we see repeatedly across this organisation?”

That shift changes everything. If students in eight classrooms are mostly completing low-demand tasks, it is unlikely that eight independent teachers all coincidentally made the same decision. The pattern may be related to curriculum materials, assessment pressure, timetable, leadership expectations, professional learning, shared beliefs about ability or the examples teachers have been given of acceptable work.

Rounds therefore create a bridge between classroom evidence and system design.

3. Begin With a Problem of Practice

A rounds visit without a problem of practice can become educational tourism. Participants see interesting classrooms, exchange impressions and leave with a long list of unrelated ideas.

A useful problem of practice is observable, high-leverage, connected to learning and within the organisation’s influence. It should be neither so broad that everything counts nor so narrow that only one teacher can affect it.

too broad: Students need to think more deeply.
better: Across subjects, what kinds of questions and tasks require students to explain, justify or revise their reasoning rather than only supply an answer?

The second question gives observers something they can actually see.

4. The Problem Must Be Owned, Not Imported

Schools can easily borrow fashionable problems. “Student agency,” “higher-order thinking,” “differentiation” and “engagement” may all be worthy ideas, but rounds become useful only when the problem is grounded in local evidence.

A school might notice that examination performance is stable but students struggle when tasks are unfamiliar. That suggests a transfer problem. Another school may see large variation between classes using the same curriculum. That suggests an implementation or task-quality problem. The problem of practice should emerge from what the system needs to understand, not from what sounds impressive in a strategy document.

5. Look at the Instructional Core

The rounds tradition emphasises the instructional core: the relationship among teacher, student and content, expressed through the work students are actually asked to do. This matters because school improvement can drift toward inputs that are easier to count than learning.

A new platform, timetable, textbook, initiative or professional-development programme matters only insofar as it changes the learning relationship. What is the teacher doing? What are students doing? What content are they working on? What cognitive demand does the task actually contain?

The task is especially important because it often predicts the ceiling of student thinking. A lesson can include energetic discussion around a task that still requires only recall.

6. Task Is Not the Worksheet; Task Is the Work Students Actually Do

A beautifully designed problem can be cognitively reduced during teaching. If the teacher supplies the key step, students may end up copying a procedure. Conversely, a simple prompt can produce demanding reasoning if students must generate, justify and compare explanations.

Rounds therefore look beyond materials to enacted task. What thinking is left for the learner? What decisions must students make? What evidence do they have to produce? How much of the intellectual work is being carried by the teacher?

7. Observation Should Be Low-Inference First

“The teacher gave too much support” is a judgement. “The teacher asked seven questions and answered five of them after less than two seconds” is a description that can support discussion about support and wait time.

Low-inference notes are not perfectly objective; observers still select what to record. But disciplined description reduces the speed at which personal preferences become system conclusions.

• Record exact question stems when possible.
• Record what students are writing or saying.
• Record the task before and after teacher intervention.
• Record who is doing the cognitive work.
• Record patterns of participation.
• Record the evidence students use.
• Record how mistakes are handled.

8. Visit Enough Classrooms to See a Pattern

One classroom is a case. Several classrooms begin to reveal a system.

Rounds typically use short visits across multiple settings rather than prolonged evaluation of one lesson. The purpose is not to understand every classroom completely. It is to collect multiple windows onto the problem of practice.

The sample still needs humility. Ten classrooms on one morning do not prove what a school always does. But recurring evidence across rooms can generate a stronger hypothesis than leadership by anecdote.

9. Separate Description, Analysis and Prediction

A disciplined debrief often moves through stages.

1. Description: What did we actually see and hear?
2. Analysis: What patterns recur across observations?
3. Prediction: If students repeatedly experience this pattern, what are they likely to know, understand or be able to do?
4. Next level of work: What should the adults learn, design or practise next?

The prediction step is powerful because it forces the team to connect adult behaviour with learner consequences. A pattern matters only because of what it is likely to produce.

10. Patterns Matter More Than Exceptional Moments

Observers naturally remember dramatic moments: an inspiring discussion, a confused class, a brilliant explanation. Rounds require a different discipline. The organisation should improve around patterns, not anecdotes.

If seven classrooms show students explaining reasoning and one does not, the single classroom may need local support. If seven show answer production without explanation, that is a system pattern. The remedy may need to involve curriculum, assessment, shared planning or professional learning rather than an individual conversation.

11. The Next Level of Work Must Be Learnable

“Improve teaching quality” is not a next level of work. “Build and rehearse question sequences that require students to justify an answer, compare two explanations and revise after counter-evidence” is closer.

The next level of work should translate a system pattern into a capability adults can develop. That may involve task design, subject knowledge, modelling, assessment literacy, questioning, feedback, student discourse or the use of representations.

Rounds without a next level of work produce insight without improvement.

12. Rounds Are Not Walkthroughs

Many schools already conduct brief classroom walkthroughs. The surface appearance can be similar: short visits, multiple rooms, notes. The mechanism is different.

A walkthrough may be used by leaders to monitor implementation or maintain visibility. Rounds are designed as a collaborative learning process around a problem of practice, with disciplined observation, collective pattern detection and a next level of work. A leader with a clipboard is not automatically conducting rounds.

13. Rounds Are Not Appraisal

The distinction is operational, not rhetorical. If teachers believe the notes will later be used to rank, reward or discipline them, the observation system changes behaviour. Teachers may stage safer lessons, conceal uncertainty and resist opening difficult classes.

Rounds need explicit rules about how notes are used, whether individual teachers are named, what information leaves the team and how formal evaluation remains separate. Developmental trust is infrastructure.

14. Rounds Are Not a Search for Best Practice

“Best practice” can encourage imitation without mechanism. An observer sees a technique in one room, copies the visible routine and later discovers that it does not work in a different subject, age group or point in the learning sequence.

Rounds are more useful when they ask what function the practice serves. What problem does this questioning routine solve? What prior knowledge does it assume? What evidence tells the teacher when to use it? What should change when the class response changes?

The aim is not to collect tricks. It is to improve collective instructional judgement.

15. Cross-Domain Comparison: Reliability Engineering

Reliability engineers do not respond to every failure as an isolated mistake. They look for recurring failure modes across components, conditions and time. A repeated pattern suggests a system cause.

Instructional rounds use a similar lens. If a learning problem appears in many rooms, the organisation should ask what shared conditions are producing it. The point is not to mechanise teaching. It is to avoid blaming individuals for patterns the system keeps reproducing.

16. Cross-Domain Comparison: Aviation Safety

High-reliability fields distinguish between the person at the sharp end and the system conditions around the work. Procedures, training, workload, equipment and communication all shape performance.

Schools benefit from the same humility. A teacher’s classroom decision matters, but the decision may be shaped by assessment expectations, curriculum pacing, shared materials or workload. Rounds create a route from visible classroom action back to invisible organisational conditions.

17. The Observation Team Needs Calibration

Different observers see different things. One notices behaviour, another language, another curriculum, another questioning. That diversity is useful, but without calibration it can become noise.

Teams should practise distinguishing description from evaluation, agree on the problem of practice, study examples of useful evidence and discuss ambiguous cases before entering classrooms. Calibration does not require identical notes. It gives the group a shared standard for what counts as evidence.

18. Student Voice Can Add Another Lens

Observation shows what students do; it does not always show what they think the task requires. Short, carefully designed student questions can help. What are you trying to learn? How will you know if your answer is strong? What do you do when you get stuck? Why did you choose that method?

Student voice should not replace observation. It triangulates it. A student may report that a task is independent while the notes show repeated teacher rescue. That mismatch is itself useful evidence.

19. Rounds Can Reveal Curriculum Problems

Suppose teachers across a school routinely compress explanation because the curriculum map is overfilled. The visible pattern may be rapid task completion and little checking for understanding. The underlying problem is not simply “teachers need better formative assessment.” It may be curriculum overload.

Good rounds do not stop at the nearest visible behaviour. They ask what system condition keeps producing it.

20. Rounds Can Reveal Assessment Problems

If common assessments reward short answer production while leaders ask teachers to develop reasoning, classrooms receive conflicting signals. Rounds may show repeated low-demand tasks because those tasks are aligned with what the system actually measures.

The next level of work may therefore sit in assessment design, not only classroom technique. This is why observation must connect back to the larger instructional system.

21. Rounds Can Reveal Professional-Learning Gaps

A school may have trained teachers in a strategy but see little evidence of it in classrooms. That does not immediately prove resistance. The training may have been conceptual rather than practical, lacked modelling, ignored subject differences or arrived without follow-up.

Rounds can show the transfer gap. Instructional coaching, rehearsal or collaborative planning may then become the next level of work.

22. The Network Effect Matters

When educators from different schools observe together, they gain a comparative frame. A practice that feels inevitable in one school becomes visible as a local choice when another school organises the work differently.

Networks can spread professional language and reduce isolation, but they also create risk. Visiting teams may import superficial solutions or compare schools competitively. The value comes from disciplined inquiry, not from touring classrooms.

23. The Evidence Base Is Useful but Not a Blank Cheque

Harvard Graduate School of Education describes instructional rounds as a collaborative process for observing, analysing and improving teaching and learning, with emphasis on the instructional core, problem of practice and collective improvement. That describes the intended mechanism.

Empirical research is smaller than the model’s visibility. A study of a two-tier instructional-rounds model in two Ontario school districts, involving teachers and leaders working on assessment for learning, reported positive changes in participants’ conceptions and implementation while also identifying challenges. Such studies are encouraging, but they do not justify claiming that rounds automatically raise achievement in every context.

The responsible position is to treat rounds as an improvement architecture whose value depends on the quality of the problem, observation, debrief and follow-through.

24. Failure Mode: The Problem of Practice Is Too Vague

“Engagement” can mean attendance, compliance, talk, curiosity, cognitive effort or persistence. If observers carry different definitions, the debrief becomes incoherent.

Define the observable phenomenon. What exactly would count as evidence? What would not?

25. Failure Mode: The Team Jumps to Solutions During the Visit

An observer sees one weak question and immediately writes “teachers need questioning training.” This is premature closure. The same surface event could come from time pressure, task design, subject knowledge, assessment expectations or a deliberate scaffold.

Rounds work better when teams earn the right to recommend by first describing the pattern and testing alternative explanations.

26. Failure Mode: The Debrief Becomes a Collection of Preferences

“I would have used group work.” “I prefer cold calling.” “I never teach it that way.” None of these statements is automatically wrong, but preference is not the same as analysis.

The debrief should return to evidence and the problem of practice. What pattern did we see? What learning consequence do we predict? What capability would change that pattern?

27. Failure Mode: Leaders Use Rounds as Surveillance

Once teachers believe rounds are a hidden compliance inspection, candour falls. The system begins observing rehearsed surfaces rather than ordinary practice.

Schools need an explicit governance rule: what data is collected, how long it is retained, whether individual names are recorded, who sees it and what decisions it can influence. Trust should not depend on vague assurances.

28. Failure Mode: No One Owns the Next Level of Work

The team identifies that tasks rarely require explanation, then everyone returns to normal work. Insight evaporates.

The next level needs an owner, timeline, practice design and evidence check. Perhaps departments redesign one task sequence, coaches rehearse facilitation moves, leaders change common assessments and the next rounds cycle checks whether cognitive demand changed.

29. A Practical Instructional-Rounds Cycle

1. Use school evidence to identify a high-leverage problem of practice.
2. Define what the problem looks like in observable terms.
3. Calibrate observers on descriptive evidence.
4. Select a reasonable range of classrooms connected to the problem.
5. Conduct short observations focused on students, teachers, content and task.
6. Record what is seen and heard without naming individual teachers in the analysis.
7. Pool evidence.
8. Identify recurring patterns.
9. Predict what students would learn if those patterns continued.
10. Generate competing explanations for the pattern.
11. Choose a next level of work that adults can actually learn and practise.
12. Assign ownership and support.
13. Implement the next level of work.
14. Return later with evidence to see whether the instructional pattern changed.

30. Example: Students Rarely Explain Answers

A school worries that students perform routine exercises but struggle on unfamiliar examination questions. The problem of practice becomes: how often do classroom tasks require learners to explain, justify, compare or revise reasoning?

Across nine rooms, observers find that teachers ask many questions, but most require one-word or numerical answers. When students hesitate, teachers often provide the missing step. Written tasks contain few prompts requiring justification. The team predicts that students may become efficient at recognising familiar routes without developing enough practice in constructing reasoning independently.

The next level of work is not “more questioning.” It is designing and rehearsing question-and-task sequences that leave explanation with the learner, followed by common review of student work. The next rounds cycle can then look for changes in enacted task.

31. Example: Feedback Is Everywhere but Revision Is Rare

Another school has invested heavily in feedback. Observers see comments, verbal prompts and correction codes in almost every class. Yet students rarely use the feedback to improve the same piece of work or a closely related task.

The system pattern is not “insufficient feedback.” It is a broken feedback-return loop. The next level of work may involve designing time and routines for response, revision and reattempt rather than training teachers to produce even more comments.

32. Missing-Node Scan: What Mature Rounds Should Test

• Is the problem of practice grounded in evidence?
• Can observers see it directly?
• Does the team know the difference between low-inference evidence and judgement?
• Are enough classrooms visited to look for a pattern?
• Is the sample broad enough to avoid one department becoming “the problem”?
• Does the debrief distinguish pattern from anecdote?
• Are predictions about student learning explicit?
• Are alternative explanations considered?
• Does the next level of work specify an adult capability?
• Are rounds separated from appraisal and ranking?
• Is there a return cycle that checks whether the pattern changed?
• Does the organisation remember what previous rounds learned?

33. Authoritative Starting Points

Harvard Graduate School of Education’s current programme page, Using Instructional Rounds to Cultivate High-Quality Learning in Schools, sets out the model’s emphasis on collaborative diagnosis, learning tasks, the instructional core and shared improvement. The foundational book is Elizabeth City, Richard Elmore, Sarah Fiarman and Lee Teitel’s Instructional Rounds in Education. For an empirical implementation example, see the study Instructional Rounds as a professional learning model for systemic implementation of Assessment for Learning.

34. The Return Path

Before rounds, the school has stories.

“Students are not independent.” “Teachers give too much help.” “The curriculum is overloaded.” “The new programme is working.”

After disciplined observation, some stories survive and some do not. The school now has a pattern it can describe. It can see what students are being asked to do, what teachers are repeatedly doing around the task and what the organisation may be producing as a result.

That pattern is not yet improvement. But it is a much better place to begin.

Instructional rounds work when a school stops trying to fix classrooms one anecdote at a time and learns to see the instructional pattern it has built together.