Teacher Development and Mentoring

Instructional Coaching: Focus, Evidence, Feedback, and Confidentiality

Published June 6, 2025 · Updated July 20, 2026

Short answer

Instructional coaching is job-embedded professional learning organized around a defined question about instruction. In a bounded cycle, a teacher and coach agree on the focus, select an instructional action, collect limited observable evidence, discuss what that evidence does and does not show, and decide whether to continue, revise, or stop. Calling an activity “coaching” does not make it voluntary, confidential, or non-evaluative; those conditions depend on the actual authority, participation, access, and data-use rules.

The research warrants a measured claim. Kraft, Blazar, and Hogan synthesized 60 studies using causal designs and reported favorable pooled estimates, but much of the evidence came from U.S. prekindergarten and elementary literacy programs, and estimates from larger effectiveness trials were smaller than those from smaller efficacy trials. The synthesis supports coaching as a professional-learning option worth specifying and testing. It does not establish that any coach, cycle, model, subject, or school will change teacher practice or student achievement.

Define the model by what people will do

Instructional coaching is a category, not a single protocol. Desimone and Pak analyze it through professional-development features such as content focus, active learning, sustained duration, coherence, and collective participation. Gibbons and Cobb likewise argue that coaching activities should be justified as opportunities for teacher learning rather than selected because they carry a coaching label. Before scheduling an observation, name the model, who controls the focus, and the decisions for which evidence may be used.

On a small screen, scroll this table horizontally to see all columns.

Instructional coaching models and adjacent roles
Approach or role	Primary purpose	Who controls the focus	Evidence and authority boundary
Teacher-directed partnership coaching	Examine a question selected by the participating teacher and test a manageable instructional adjustment.	The teacher selects or accepts the focus and chooses among feasible next steps.	Notes support the agreed conversation. They are not evaluation records unless that different use was disclosed and authorized in advance.
Content-focused coaching	Develop subject-specific planning, task analysis, enactment, and review around a defined curriculum or pedagogy.	The program may define the content area; the agreement should still state what the teacher can choose.	Evidence is interpreted against the particular content model. Findings from one literacy, mathematics, or science model should not be generalized to another.
Implementation or fidelity coaching	Support accurate use of a specified practice after initial training, often through modeling, observation, feedback, and rehearsal.	The required practice may be program-defined; the teacher and coach can define the immediate implementation question.	An implementation measure indicates whether specified components were observed. It does not by itself establish quality, appropriateness, or student learning.
Student-work inquiry coaching	Use a defined sample of student responses or work to examine the alignment among the task, instruction, and success criteria.	The teacher and coach agree on the sample and the instructional question before viewing the work.	Student work can show features present in that sample. It does not reveal motivation, diagnosis, or the cause of a pattern.
Mentoring	Support induction, role navigation, professional judgment, and access to institutional knowledge, often for a new teacher.	The mentee’s needs and induction requirements can both shape the agenda.	Mentoring may contain a coaching cycle, but its broader career and organizational remit is different from a bounded instructional question.
Supervision or formal evaluation	Monitor requirements, judge performance against formal criteria, or make employment and accountability decisions.	Policy and the authorized evaluator determine criteria and required evidence.	The evaluator must follow the applicable process. A person with this authority should not present an evaluative observation as confidential coaching.

Peer observation is an observation arrangement rather than a complete coaching model. It can supply evidence within a cycle, operate as reciprocal professional learning, or form part of evaluation. The agreement and use of records determine which function it serves.

Set consent, confidentiality, and role boundaries first

Use a short written agreement before collecting notes, student work, audio, or video. If participation is required, state that directly; required participation should not be described as voluntary. Separate permissions may still be needed for recording, access to student information, or any use beyond ordinary coaching. Apply local contracts, collective-bargaining provisions, records rules, safeguarding duties, and school or district policy.

Participation: Record whether the program and this cycle are voluntary or required, who selected the focus, and which parts the teacher may decline or pause.
Purpose: Name the instructional question and exclude unrelated monitoring. A broad instruction to “watch the class” is not a bounded focus.
Access: List who may see raw notes, recordings, coded tallies, student work, and the cycle summary. Do not rely on an undefined promise that data will remain “between us.”
Confidentiality limits: State required exceptions before the cycle, including applicable safety, safeguarding, harassment, discrimination, legal, and mandatory-reporting routes. Promise only the confidentiality the program can maintain.
Evaluation separation: State whether the coach has supervisory authority and whether any coaching record can enter evaluation or personnel processes. If it can, do not call the record non-evaluative.
Student privacy: Minimize student-identifiable information, use approved systems, and obtain any permission required for recording. An invitation to observe a teacher is not permission to create or reuse a recording of students.
Retention: Name where records are stored, how long they are kept, who deletes them, and what aggregate information the program will retain.
Withdrawal and concern route: Explain how a teacher can pause a voluntary cycle, challenge a record, report coercion or retaliation, and contact the program lead, administrator, human-resources office, or representative specified by local procedure.

Write an observable evidence plan

Begin with a question that can be answered from a defined lesson segment or set of work. “How engaged were students?” invites inference. “During the first ten minutes of independent practice, how many students began the assigned item without an additional direction, and which direction preceded the start?” names an event, a window, and a source. Counts do not explain why an event occurred, so the debrief must keep observation separate from interpretation.

A usable evidence plan contains the following fields:

Question: One sentence about an instructional action or a feature of student work.
Indicator: A behavior, sequence, duration, count, quotation, or work feature that an observer can record without assigning motive.
Sample: The class, lesson segment, task, or set of work included—and what is excluded.
Collection method: Tally, time stamp, seating-map code, transcript excerpt, implementation checklist, or pre-agreed work-sample review.
Baseline or comparison: The prior observation or work sample, if one is needed and reasonably comparable.
Implementation check: The evidence that will show whether the planned action actually occurred before interpreting any other pattern.
Privacy controls: Coded identifiers, approved storage, access, retention, and deletion date.
Decision rule: What evidence would lead to repeat, revise, seek specialist input, or stop.
Limit statement: What this evidence cannot establish, including causation, diagnosis, motivation, or general performance.

Useful sources can include a sequence of teacher questions, wait-time samples, opportunities to respond, transition duration, an implementation checklist, or a defined feature in student work. Teacher reflection and a usefulness survey are also legitimate evidence about the coaching experience. They should not be reported as proof that instruction or student learning changed.

Coaching-cycle decision table

The cycle should produce a decision at each stage rather than accumulating observations without a stated use.

On a small screen, scroll this table horizontally to see all columns.

Instructional coaching cycle stages, evidence, and decisions
Stage	Evidence available	Decision question	Possible route
1. Contract	Written participation, role, access, confidentiality, retention, and escalation terms.	Can the stated non-evaluative boundary actually be honored?	Proceed; revise the agreement; transfer to an independent coach; or stop before collecting data.
2. Focus and baseline	Teacher-selected question, context, existing lesson material, and a limited baseline when needed.	Is the question observable and within the coach’s competence and role?	Keep the focus; narrow it; define a different indicator; or refer a content, access, safety, or policy question.
3. Plan and rehearse	Selected action, rationale, success criteria, rehearsal notes, and implementation check.	Is the action feasible, instructionally aligned, and ready to observe?	Teach as planned; model or rehearse; adapt for access; or obtain specialist review.
4. Observe and collect	Only the agreed codes, time stamps, work sample, or checklist.	Was the plan implemented, and was the evidence collected as agreed?	Debrief; repeat because implementation or collection was incomplete; or discard data gathered outside scope.
5. Debrief	Teacher account, descriptive record, implementation evidence, and relevant context.	Which interpretations fit the record, and what remains uncertain?	Repeat without change; revise one element; gather a different source; seek specialist input; or end the cycle.
6. Follow-up	A second comparable sample when appropriate, plus the teacher’s review of feasibility and usefulness.	Is there enough relevant evidence for the next bounded decision?	Continue for another specified cycle; retain the action provisionally; revise; or stop. Do not convert a local pattern into a general outcome claim.
Pause or close	Withdrawal, role conflict, confidentiality breach, unsafe practice, policy concern, or a question beyond coaching scope.	Which authorized process now applies?	Secure or delete records as policy requires, notify the designated contact, and document why coaching ended without conducting an independent investigation.

Worked example: discussion participation

This is a hypothetical planning example; the counts are invented and do not describe a real class. A middle-school social studies teacher selects this question: “During the first 12 minutes of source-based discussion, how many students offer at least one claim linked to a source, and which prompt immediately preceded the response?” The question concerns an observable distribution of contributions. It does not treat speaking frequency as a measure of engagement, understanding, confidence, or achievement.

Agreement and baseline

The cycle is non-evaluative and voluntary. The coach has no supervisory role.
The coach uses roster codes, records prompt type and source-linked contributions, and does not record student names in the coaching file.
The teacher and coach can see the tally. The administrator receives only an aggregate notice that a cycle was completed. The coded sheet is deleted after the follow-up meeting.
In the baseline segment, 8 students offer a claim linked to a source. The coach also records that the teacher moves from the opening question directly to whole-group responses.

Action and observation

The teacher chooses one adjustment: students first draft a claim from one of two provided sources, then rehearse it with a partner before the whole-group discussion. The implementation check is limited to whether drafting time, source choice, and partner rehearsal occurred. During the follow-up segment, the coach records that the planned steps occurred and that 13 students offered a source-linked claim.

Debrief and next decision

The two tallies show that the distribution differed across two lesson segments. They do not show that the routine caused the difference, that every student had an accessible participation route, or that historical reasoning improved. The teacher decides to repeat the routine once with the same observation window and add a small, preselected sample of written claims to check whether students cited a source accurately. If the speaking tally and written sample point in different directions, the teacher and coach will revise the question rather than select the more favorable measure.

Conduct a descriptive feedback conversation

Teacher account first: Ask what the teacher noticed and which parts of the plan were enacted or altered.
Implementation check: Confirm whether the agreed action occurred before discussing other evidence.
Description: Share counts, sequences, quotations, time stamps, or work features without assigning intent.
Interpretation: Generate more than one plausible explanation and identify relevant context that was not observed.
Boundary: State what the sample cannot answer and whether additional information would be proportionate.
Decision: Record one next action, the evidence for reviewing it, the responsible person, and the follow-up date.

Replace ratings such as “strong questioning” with a record such as “the teacher asked four open questions and provided between two and six seconds before calling on a respondent.” Replace “students understood” with the feature present in the agreed work sample. Descriptive evidence is still selective; the teacher should be able to correct factual errors and add context.

Implementation sequence for a school or district

Publish a coaching charter. Define the purpose, chosen model, participation status, authority lines, confidentiality limits, data use, complaint route, and relationship to evaluation.
Map role conflicts. Identify every coach who also supervises, evaluates, assigns work, or reports on personnel. Separate roles where possible; otherwise disclose the conflict and avoid claiming confidentiality that cannot be maintained.
Specify coach work. List the activities the model permits—planning, content analysis, modeling, rehearsal, observation, work review, or facilitation—and the activities outside scope.
Prepare coaches. Provide subject or pedagogical preparation relevant to the assignment, practice in descriptive data collection and feedback, and training on privacy, records, access, safeguarding, difficult conversations, and referral routes.
Protect capacity. Set a feasible caseload, meeting time, observation time, and record-handling process. A calendar filled with brief contacts should not be reported as completed cycles.
Pilot the operating rules. Begin with a defined group and review whether contracts, time, evidence plans, debriefs, and deletion rules were followed. Teacher reports of usefulness can guide process revision but do not establish instructional effects.
Review aggregate process evidence. Track completed stages, delays, withdrawals, boundary concerns, coach workload, and adherence to data rules. Keep identifiable coaching records out of program reports unless access and purpose were authorized.
Decide on continuation. Continue, revise, narrow, or stop the program based on implementation evidence, capacity, participation conditions, and independent evaluation where an outcome claim is required. Do not infer impact from the number of meetings held.

Escalation and stop limits

A coach can document an observation, ask questions, support planning, and connect a teacher with an authorized resource. A coach should not diagnose a student or colleague, investigate alleged misconduct, decide disability services, provide mental-health treatment, promise legal confidentiality, or conduct an undisclosed personnel evaluation.

Immediate safety or safeguarding concern: Follow the school or district reporting procedure promptly. Do not keep the matter inside a coaching agreement.
Harassment, discrimination, retaliation, or employment dispute: Explain the confidentiality limit and use the designated administrator, human-resources, civil-rights, union or employee-representation, or complaint route. The coach does not independently determine the allegation.
Student access, disability, language-service, or health need: Bring the task and observable barrier to the responsible student-support team or qualified staff. Coaching notes do not replace an individualized plan or authorized decision.
Formal performance concern: Use the established supervision or evaluation process. Do not repurpose confidential coaching notes unless the disclosed policy and applicable rules authorize that use.
Question beyond coach expertise: Pause the recommendation and consult a qualified content specialist, program lead, accessibility contact, data-privacy contact, or administrator.
Coercion, role confusion, or confidentiality breach: Stop collection, secure the record, notify the designated program contact, and decide whether the cycle can resume with a different coach or corrected agreement.

Pre-cycle checklist

The coaching model and permitted activities are named.
The teacher selected or explicitly accepted one observable instructional focus.
Voluntary or required participation is stated accurately.
Coach, mentor, supervisor, evaluator, administrator, and data-access roles are not conflated.
Raw-record access, aggregate reporting, storage, retention, and deletion are written down.
Confidentiality exceptions and the concern route were explained before collection.
The sample and observation window are limited.
The implementation check is separate from the practice or work evidence.
Student-identifiable information is minimized and any recording permission is confirmed.
The decision rule includes continue, revise, refer, and stop options.
The plan states what the evidence cannot establish.
The feedback meeting is scheduled before the observation occurs.

Sources and further reading

Kraft, Blazar, and Hogan, “The Effect of Teacher Coaching on Instruction and Achievement: A Meta-Analysis of the Causal Evidence” (Review of Educational Research, 2018) — Used for the broad research boundary: 60 causal-design studies, concentration in U.S. early literacy settings, and reduced estimates in larger effectiveness trials.
Desimone and Pak, “Instructional Coaching as High-Quality Professional Development” (Theory Into Practice, 2017) — Used for the professional-development features of content focus, active learning, sustained duration, coherence, and collective participation, and for the authors’ caution about the empirical base available at publication.
Gibbons and Cobb, “Focusing on Teacher Learning Opportunities to Identify Potentially Productive Coaching Activities” (Journal of Teacher Education, 2017) — Used to frame coaching activities as deliberately designed opportunities for teacher learning; this is a conceptual analysis focused on mathematics and science rather than a general outcome study.
Kretlow and Bartholomew, “Using Coaching to Improve the Fidelity of Evidence-Based Practices: A Review of Studies” (Teacher Education and Special Education, 2010) — Used for the bounded implementation sequence of initial training, observation, feedback, and modeling; the review covered 13 studies and particular practices, so it is not treated as a universal coaching protocol.
Matsumura, Garnier, and Spybrook, “The Effect of Content-Focused Coaching on the Quality of Classroom Text Discussions” (Journal of Teacher Education, 2012) — Used to illustrate that a studied coaching model can combine planning, enactment, reflection, pedagogical content, and coach support; findings concern that elementary literacy program and its measures.
Regional Educational Laboratory Pacific, “Improving Teacher Performance Through Instructional Coaching” — Used as an official practice overview of coaching forms, recurring-cycle elements, and role descriptions; it is a secondary summary, not evidence that a local program will produce a stated outcome.

Research findings depend on the coaching model, content, participants, implementation, comparison, and measure. Local process evidence can guide a local coaching decision; claims about teacher or student outcomes require an evaluation designed for that claim.