The Tutor Handbook · Volume 0186 · Series ID THB-0186
The Tutor Handbook: Complete Series Index
A beautifully prepared lesson can still be badly taught
A tutor arrives with everything ready. The worked examples are sequenced. The worksheet has three levels. The extension question is printed. The recap is timed. There is a model answer for every task and a contingency slide for “common misconceptions”. Nothing has been left to chance.
Ten minutes into the lesson, the learner gives an answer that does not fit the predicted misconception. The tutor notices, but the plan is strong and the clock is moving. So the tutor continues. The learner completes the planned scaffold, reaches the expected answer and appears to recover. Later, on a fresh question, the same underlying confusion reappears because the lesson repaired the misconception the tutor prepared for, not the misconception the learner actually had.
The opposite tutor arrives with almost nothing prepared because “good teaching is responsive”. Every example is invented live. The lesson follows learner questions wherever they lead. Sometimes this produces a rich conversation. Sometimes it produces missing materials, poorly chosen tasks, wasted minutes and explanations built under unnecessary pressure.
Both tutors misunderstand planning. Preparation should reduce avoidable uncertainty so the tutor has more attention available for the uncertainty that only live teaching can reveal.
The Preparation–Contingency Boundary asks what a tutor should decide before the lesson and what must remain deliberately open until learner evidence arrives.
Quick answer
Prepare the stable parts: learning target, relevant prerequisites, likely evidence sources, accurate examples, task progression, legitimate access supports, material logistics, important misconceptions worth watching, and the boundaries around help. Do not pre-decide the learner’s diagnosis, the exact amount of scaffolding, whether the group will need to branch, how long a repair will take, or whether the planned extension will be appropriate.
A strong plan contains decision points, not only activities. It says, in effect: “If the learner demonstrates X independently, move here. If the evidence suggests Y, check this before teaching. If the group diverges, use this temporary branch. If the result is ambiguous, do not force the lesson to finish the planned chapter.”
The tutor should prepare enough that live adaptation is informed rather than improvised, while leaving enough open that the learner’s actual state can still change the route.
Why current apex guidance makes this gap visible
The National Student Support Accelerator’s current Toolkit for Tutoring Programs recommends preparing necessary materials in advance, using data-informed session goals, establishing consistent session structures and employing facilitation moves that respond to learner thinking. This combination is important. High-quality tutoring is not “winging it”, but neither is it delivery of a fixed script regardless of evidence.
AERO’s current Scaffold Practice guidance makes a related distinction between planned scaffolding, designed during preparation, and contingent scaffolding, introduced during lessons in response to learning needs as they arise. The guide also emphasises fading scaffolds as proficiency develops. That is school-teaching guidance rather than a tested private-tuition planning protocol, but the distinction is directly useful: some supports can be anticipated, while others should depend on what the learner actually shows.
The OECD Teaching Compass frames teachers as adaptive experts and co-designers of learning rather than passive implementers. That policy framework supports the professional idea behind this boundary: planning and adaptation should work together.
The Tutor Handbook already owns session structure, scaffolding, tutor-side checks and route changes in other volumes. This article owns a narrower planning question: what belongs in the plan before the learner arrives, and what must the tutor refuse to decide too early?
Planning should remove preventable uncertainty
Some uncertainty has no educational value. The tutor should not discover during the lesson that the worksheet is missing, the graph is unreadable, the answer key is wrong, the online link requires a login, or the selected question depends on a prerequisite the group has never learned.
Preparation should remove those avoidable failures. Check materials. Solve the problems. Inspect the language. Verify units, diagrams and answer paths. Decide what the main learning target is. Know the school context where it matters. Identify which supports are legitimate access supports and which supports would carry the target cognition.
Every minute saved from preventable logistics becomes attention the tutor can spend on learner evidence.
This is the real value of preparation. It is not to predict the learner perfectly. It is to free the tutor from predictable noise.
Planning should not manufacture certainty about the learner
The dangerous part of planning is not being organised. It is writing a story about the learner before the evidence arrives.
The tutor sees a past error and writes, “Learner does not understand fractions.” They prepare twenty fraction questions. The next lesson reveals that the learner’s concept is secure but they misread one mixed-number instruction. The plan becomes a machine for confirming an outdated diagnosis.
A better note is provisional: “Previous work contained an error converting mixed numbers. First check whether the issue is conceptual, procedural or task-reading before deciding the repair.”
The difference in wording matters. One plan contains a conclusion. The other contains a question.
Good planning preserves hypotheses as hypotheses.
The plan should contain a learner decision, not only a tutor action
Traditional lesson plans often describe what the tutor will do: explain, model, give worksheet, review answer, assign homework. That sequence can run perfectly while the tutor collects little information about whether learners are ready to move.
A more useful plan includes what evidence the learner will produce at important moments. “After modelling, each learner attempts one fresh item without method cues.” “Before extension, learner explains which condition distinguishes the two cases.” “After feedback, learner revises one new paragraph rather than only correcting the marked one.”
These evidence points create branching opportunities. The tutor does not need to predict every response. They need to know which responses would change the next move.
Planning becomes a set of conditional decisions rather than a list of activities.
Composite case: Beatrice and the perfect misconception slide
The following case is fictional and constructed for teaching. Beatrice’s Science tutor prepares a lesson on heat transfer. The tutor expects Beatrice to confuse temperature with heat, a common difficulty, and prepares an excellent comparison diagram.
Early in the lesson, Beatrice correctly distinguishes temperature and thermal energy but repeatedly assumes that metal “contains coldness” because it feels colder to touch. The prepared misconception is not the live one.
A rigid plan delivers the temperature-versus-heat slide anyway. Beatrice answers correctly. The tutor marks the misconception as repaired. The lesson proceeds.
A contingent plan uses the prepared slide only if the evidence calls for it. Instead, the tutor pauses and creates a short comparison involving different materials at the same room temperature, asking Beatrice to explain the sensation and energy transfer. The prepared material remains unused.
That is not wasted preparation. The tutor’s preparation created enough conceptual clarity to recognise that the wrong misconception had appeared.
Unused preparation is not failure
Tutors sometimes feel pressure to “use what they prepared”. They spent time creating an activity, so leaving it out feels inefficient. This is sunk-cost thinking inside a lesson.
Prepared material is an option, not a debt the learner owes the tutor. If the learner already demonstrates the target, the tutor can skip the scaffold. If a different weakness appears, the tutor can postpone the planned extension. If the group needs more independent practice than expected, the discussion activity can wait.
The success of preparation should be measured by whether it improved the tutor’s readiness to make good decisions, not by the percentage of slides shown.
A teacher who abandons the right activity for the right reason has used the plan professionally.
Plan the invariant; adapt the surface
A useful planning habit is to separate what must remain stable from what can change. The invariant may be the learning objective, the evidence needed before progression, the rule that all three learners produce an independent first attempt, or the boundary that the tutor will not author assessed work.
The surface can vary: examples, pacing, representations, order of tasks, amount of practice, grouping moves and the exact language of prompts.
This resembles the Instructional-Practice Consistency Gate but operates within one tutor’s lesson preparation. The plan should protect the mechanism while leaving room to change the route.
If the tutor does not know which parts are invariant, adaptation becomes risky because any change might remove the thing that made the task educationally useful.
Prepare examples as a family, not a fixed sequence
One way to support contingent teaching is to prepare an example family rather than one linear worksheet. Include an easier diagnostic example, a standard example, a contrast case, a common-error case and a transfer example. The tutor can then select based on live evidence.
This is different from dumping five levels of worksheet onto the learner. The tutor knows what each example is for. If the learner’s first response is strong, the easier item may never be used. If the learner’s error is ambiguous, the contrast case can discriminate. If performance is fluent but narrow, the transfer item becomes useful.
The preparation pays off because the tutor has choices that are accurate and purposeful under time pressure.
The tutor should still resist turning the example family into a decision tree so complex that live teaching becomes form-filling.
Prepare likely misconceptions, but do not hunt for them
Anticipating misconceptions helps tutors recognise patterns. It also creates confirmation bias. Once a tutor expects a particular error, ambiguous learner behaviour can be forced into that category.
A good misconception note includes a falsifier. “Possible sign-distribution confusion. Check with an item where the sign changes but the algebraic structure remains simple. If learner handles that accurately, do not keep teaching sign rules.”
This makes the plan diagnostic rather than deterministic.
The Tutor-Side Check is relevant here: if several learners show the same unexpected confusion, inspect the explanation, material or sequence before assuming identical learner deficits. Preparation should make the tutor more open to alternative causes, not less.
Plan support levels, not support delivery dates
A tutor can prepare several scaffold levels: full worked model, partial worked example, prompt, checklist, visual representation or simply extra wait time. What should remain contingent is which level the learner needs and when it should fade.
A fixed schedule such as “use the organiser for three weeks, then remove it” is convenient but may ignore learner evidence. AERO’s Scaffold Practice guidance explicitly supports planned and contingent scaffolds and gradual removal as proficiency develops.
The tutor can therefore plan the scaffold architecture before the lesson while making activation conditional. “If the learner cannot identify the first decision independently, offer representation A. If they can identify it but execution fails, use prompt B. After successful supported practice, test a fresh item with one layer removed.”
This is prepared responsiveness rather than improvisation.
Composite case: a three-learner English group diverges
This case is fictional. Alicia, Ciara and Denise are preparing for situational writing. The tutor plans a common mini-lesson on audience and purpose, followed by one independent task.
During the first response, Alicia identifies audience and purpose correctly but writes too broadly. Ciara misunderstands the required action entirely. Denise identifies the task accurately but lacks vocabulary to express the intended tone.
A rigid plan treats the class as one misconception and teaches the prepared mini-lesson in full. A completely improvised tutor splits into three unrelated private lessons.
A stronger tutor uses the prepared common frame, then branches briefly. Ciara receives a narrower task-interpretation check. Denise receives a language-choice contrast. Alicia starts the independent response with one constraint added. Ten minutes later, the tutor reconnects the group around comparing how audience changes wording.
The plan survived, but not as choreography. Its invariant survived: learners must interpret audience and purpose before drafting. The exact support became contingent.
Preparation should include what you will not do
Boundaries are easier to preserve when they are decided before pressure arrives. The tutor can precommit that they will not write the learner’s assessed response, will not diagnose a clinical condition, will not introduce an unverified AI-generated question during live teaching, and will not remove a legitimate access support merely to make a check feel more independent.
These “do not” decisions reduce moral and cognitive load in the moment. When a parent sends urgent school work five minutes before the lesson, the tutor already knows the Homework Collision Gate rather than improvising from guilt. When the learner asks for the exact sentence to submit, the Submission Boundary is already clear.
Preparation is therefore partly about professional governance, not only pedagogy.
Do not over-plan time to the minute
Minute-by-minute plans can help novices estimate lesson shape, but false precision can damage responsiveness. A planned seven-minute discussion may need three minutes because the learners already understand. A planned ten-minute practice may need twenty because the first independent attempt exposes a load-bearing gap.
Time budgets are useful when they express priority rather than destiny. “Protect at least fifteen minutes for independent application” is often more educationally meaningful than assigning every transition a timestamp.
The tutor should know which part of the lesson can compress safely and which part cannot be sacrificed. If the lesson runs long, what gets dropped first? If diagnosis takes longer than expected, will extension disappear, or will independent practice be squeezed instead?
Planning for trade-offs is more useful than pretending trade-offs will not occur.
Contingency is not permission to chase every learner question
Responsive teaching can drift into endless detours. A learner asks an interesting question. The tutor follows it. Another learner adds a connection. Soon the lesson is intellectually rich but far from the target, and the prerequisite that brought the learners to tuition remains unresolved.
Contingent teaching still needs scope. A tutor can acknowledge a valuable question and park it. Frontier exploration may be appropriate when the learner is ready and the time budget supports it. In Repair mode, the question may be fascinating but costly.
The plan provides a reference point for judging detours. Responsiveness means changing the route because evidence or value justifies it, not following every available branch.
The plan should include a recovery route
Tutors often prepare the ideal lesson but not what happens when the lesson fails. A stronger plan includes one or two recovery options. If the explanation does not land, what alternate representation is available? If the group cannot begin the independent task, what smaller diagnostic check will locate the missing prerequisite? If the technology fails, is the learning job still possible offline?
Recovery planning is different from predicting every problem. It identifies likely single points of failure and ensures the tutor is not forced into answer-giving simply because the intended route broke.
This is especially valuable for newer tutors. Experienced tutors carry many recovery patterns in memory. Novices benefit from having some prepared explicitly until their professional repertoire grows.
Prepare evidence collection lightly
A tutor can also over-plan measurement. Every task becomes a data point, every error a code, every learner response a field to record. The Evidence-Capture Burden Gate already protects against documentation taking over teaching.
For preparation, the rule is simple: decide which one or two pieces of evidence are likely to change the route. Those deserve deliberate attention. Everything else can remain ordinary teaching information unless a new concern emerges.
For example, if today’s decision is whether a scaffold can fade, the tutor needs evidence from a fresh attempt with reduced support. They do not need a full dashboard of every question answered during the lesson.
Planning should sharpen observation, not create administrative fog.
The difference between preparedness and prediction
Preparedness means the tutor has accurate materials, knows the target, understands likely dependencies, can recognise important error patterns and has safe alternatives available. Prediction means the tutor believes they know exactly what the learner will do.
The first is professional. The second should remain modest.
Even with years of experience, tutors will meet surprising responses. Expertise should improve the speed and quality of interpretation, not create the illusion that surprise has disappeared. A strong plan expects uncertainty as part of teaching.
This attitude also protects learner dignity. The child arrives as a person who can produce new evidence, not as the tutor’s prewritten case study.
Planning for three learners requires common and individual layers
Small-group preparation works best when the tutor identifies what can be shared and what may need individual checks. The common layer includes the main concept, representations, core task and shared discussion. The individual layer includes known fragilities, legitimate access supports and one or two learner-specific decision points.
The tutor should avoid writing three entirely separate lesson plans unless the group is already structurally broken. If every learner requires unrelated objectives, the Grouping or Regrouping Gate may be more relevant than better preparation.
Likewise, the tutor should avoid one undifferentiated plan that assumes all three learners will move together. Good preparation acknowledges likely variation without building three parallel classrooms.
The plan should make branching possible and reconnection likely.
The learner can sometimes be part of contingency planning
Older learners can help identify where uncertainty lies. At the start of a lesson, a tutor may ask which recent work felt unstable, which school topic is arriving next, or what happened when the learner attempted the previous plan independently. This is not outsourcing diagnosis to self-report. It adds context before the tutor checks directly.
Learner agency can also shape extension choices. If two mathematically valid challenge routes exist, a learner may choose the context that interests them while the tutor preserves the underlying target.
The key is that learner preference informs the plan without becoming the sole evidence for what the learner needs.
Preparation for online tutoring needs another layer
Live-online tuition adds predictable technical dependencies: shared workspace, camera or audio quality, document visibility, login access and how the tutor sees the learner’s working. These should be prepared because technical friction can consume attention rapidly.
At the same time, the tutor should not assume an online lesson can reproduce the in-person contingency repertoire. A physical manipulative may need a digital substitute. Monitoring three learners’ written work may require a shared document routine. Turn-taking may need more explicit structure.
The Modality-Shift Gate owns the broader transition. The preparation boundary adds a practical rule: prepare the infrastructure differences before the lesson so live adaptation remains about learning rather than avoidable platform confusion.
When AI enters preparation
AI can generate examples, rewrite explanations and suggest questions quickly. That makes pre-lesson preparation cheaper and creates a new verification burden. A tutor should not discover an AI-generated error while a learner is already working through it.
If AI assists preparation, the tutor remains responsible for checking mathematical correctness, source claims, ambiguity, answer keys, age appropriateness and whether the task matches the intended capability. The tutor should also know which parts came from a generic system and which parts are grounded in the learner’s actual evidence.
AI is useful for generating options. It should not pre-decide the learner’s diagnosis from thin notes or create a false sense that every contingency has been simulated.
The Human-Supported AI Engagement Gate covers learner-facing use. Here, the rule is tutor-facing: generated preparation still needs professional ownership.
Coaching should inspect planning decisions, not only the final plan
A coach who reviews lesson plans can easily reward neatness. The document is complete, objectives are written, timings add up, resources are attached. Yet the plan may still pre-decide the learner.
A stronger coaching conversation asks: Which parts are fixed? Which decisions are waiting for evidence? What would make you skip this task? What would make you increase support? Which response would challenge your current hypothesis? What is your recovery route if the first explanation fails?
These questions reveal whether the tutor is planning adaptively.
For experienced tutors, the written plan may become shorter because much of the structure is internalised. The evidence standard should not become “more paper equals better planning”. The coach needs enough visibility to understand the tutor’s decision architecture, not an administrative performance.
Failure modes
The completion trap: the tutor tries to use every prepared activity because preparation effort creates sunk cost.
The diagnosis-before-evidence trap: the lesson is designed to confirm a learner story formed from old data.
The improvisation-romance trap: lack of preparation is reframed as responsiveness.
The decision-tree trap: planning becomes so complicated that the tutor follows branches mechanically instead of observing the learner.
The minute-by-minute trap: timing precision outranks learning evidence.
The misconception-hunt trap: tutors search for predicted errors and miss the one actually present.
The scaffold calendar trap: support is introduced and removed on dates rather than on evidence.
The resource dump: tutors prepare multiple levels but cannot explain what decision each resource supports.
The AI-option trap: generated tasks create apparent abundance without verification or coherent purpose.
The detour trap: every learner question is treated as evidence that the lesson should change direction.
Evidence boundaries and sources
The National Student Support Accelerator’s Strengthening Instructional Practices and Establishing Routines and Structures recommends preparing materials in advance, setting clear data-informed session goals and using facilitation practices responsive to student thinking. These are programme-quality recommendations and do not establish one optimal lesson-plan format.
AERO’s Scaffold Practice guide, published 29 November 2024 and updated 14 May 2026, distinguishes planned scaffolding from contingent scaffolding and recommends fading support as learners develop proficiency. It is school-oriented research-informed guidance rather than a trial of this exact tutoring protocol.
AERO’s Monitor Progress guide supports frequent checks for understanding and responsive additional instruction, guidance or feedback. The OECD Teaching Compass, published 30 May 2025, frames teachers as adaptive experts and agents of change. These sources support planning that leaves room for evidence-led adaptation while preserving professional judgement.
Parents: what preparation should look like from the outside
A prepared tutor should not need to spend lesson time finding basic materials, solving the worksheet for the first time or deciding what the topic is. Parents are right to expect organisation.
But a good tutor may also change the plan. The learner may skip work because they are already secure, spend longer on one prerequisite, receive a different example or postpone extension. That flexibility is not necessarily poor planning. It may be the visible result of better planning: the tutor knew what could change and why.
The useful question is not, “Did you finish everything you planned?” It is, “What did the learner evidence change, and what does that mean for the next step?”
The end state
Preparation should make a tutor calmer, more accurate and more available to observe. It should not make the learner’s future less visible because the tutor has already written it down.
The stable parts of teaching deserve preparation: purpose, materials, evidence points, safe boundaries, likely dependencies and recovery options. The uncertain parts deserve professional openness: the learner’s exact misconception, the amount of help required, the pace, the branch, the moment a scaffold can fade and whether the planned extension still makes sense.
The best plan therefore contains both confidence and humility. It says, “I know what this lesson is for. I know what evidence matters. I am ready with several good options. I do not yet know which option this learner will earn.”
That is the Preparation–Contingency Boundary. Plan the lesson. Do not pre-write the learner.
A final planning test: could another tutor understand the decision without copying the lesson?
One useful test of a strong plan is portability. If another competent tutor had to cover the session unexpectedly, could they understand the target, the known evidence, the important boundaries and the decisions that are deliberately waiting for live information? They should not need to imitate the original tutor’s personality or every planned sentence. They should be able to preserve the educational route.
This makes the plan different from a performance script. A script tells another tutor what to say. A portable professional plan tells them what the lesson is trying to establish, which prior knowledge matters, what support is currently legitimate, what evidence would justify progression and what unresolved question should not be assumed away.
Portability also exposes over-planning. If the document is so dense that the covering tutor cannot identify the central decision, the plan has accumulated detail without increasing usability. Conversely, if the note says only “continue algebra worksheet”, the route depends too heavily on tacit memory.
The right level is the smallest sufficient preparation that keeps the learner’s route coherent while leaving live judgement alive. That is a useful standard for ordinary tutors, coaches and programme leaders because it connects preparation to continuity rather than paperwork.
Preparation quality is visible in what the tutor notices, not only in what they bring
A final paradox of planning is that the most valuable preparation may disappear from view during the lesson. The tutor has studied the concept deeply enough to recognise a surprising misconception. They have solved the tasks carefully enough to notice that a learner’s alternative method is valid. They have reviewed the learner’s recent work closely enough to know that today’s hesitation is new rather than habitual. None of this produces a visible worksheet or slide, yet it changes the quality of live judgement.
This is why programme leaders should be cautious about measuring preparation through artefact volume. A large folder of resources may indicate diligence, but it can also indicate duplication, uncertainty or a belief that every possible branch must be prebuilt. A shorter plan can be stronger when it reflects secure subject knowledge, a clear learning purpose and a small number of decision-relevant contingencies.
Coaches can inspect preparation by asking the tutor to reconstruct one live choice. What did you expect before the lesson? What evidence changed that expectation? Which prepared option did you use or discard? What part of the plan remained invariant? What will you prepare differently next time because of what you learned? This conversation turns planning into professional learning rather than document compliance.
It also creates a healthy feedback loop. Live contingency should improve future preparation. If the same unexpected problem appears across several learners, it may no longer be unexpected and should enter the planned repertoire. If a prepared scaffold is almost never needed, it can be retired or moved farther from the default route. If tutors repeatedly improvise the same useful contrast case, the programme can vet it and add it to shared materials.
Preparation and contingency therefore educate each other. Planning makes adaptation safer; adaptation makes future planning wiser. The boundary is not a wall between the two. It is a disciplined exchange between what can be known before the lesson and what only the learner can reveal once the lesson begins.
