The Tutor Handbook Vol No.0054 | The Drift Check — How a Tutor Detects When an Adopted Learning Routine Quietly Stops Being the Routine That Was Chosen

The Tutor Handbook · Volume 0054 · Series ID THB-0054

The new routine worked.

Three weeks later, it is still there.

At least, everybody thinks it is.

Alicia’s adopted Mathematics check was simple: on high-risk multi-step work, scan the sign, unit and copied value before moving on. During the trial, she used all three checks. A month later she still says she is “doing the check”, but in practice she glances only at the final answer.

Beatrice’s adopted writing rule was equally clear: decide what the paragraph is establishing, then decide what evidence or development makes the idea believable. At first she used the rule before drafting. Later, under school pressure, she fills in the paragraph first and retrofits a planning sentence afterward.

Denise’s timed Additional Mathematics routine began as one short weekly performance block. Then examinations came closer. One block became two. Two became most of the lesson. Eventually even new learning is being rushed under the clock.

Faith’s help gate began with a sensible independence rule: state the task, try one plausible first move, then ask a specific question if the route fails. Over time, “try one plausible first move” quietly becomes “struggle for five minutes before you are allowed to ask”. The surface still resembles the original routine. The educational mechanism has changed.

A routine can remain visible while its function quietly drifts.

This is one of the least dramatic and most consequential problems in long-running tutoring.

A poor intervention is easy to notice when it fails immediately. A drifting intervention is harder. It may still look familiar. The worksheet is still there. The checklist still has the same name. The weekly review still happens on Sunday. The tutor still says the same phrase. The learner may even still be improving.

But the lived routine is no longer quite the routine that crossed the Adoption Gate.

The Drift Check is the tutor’s periodic comparison between an adopted learning routine and the routine the learner is actually living, so that useful adaptation can be preserved while accidental change, support creep, shortcut accumulation and purpose loss are detected before they become the new normal.

The previous volume, The Tutor Handbook Vol No.0053 | The Adoption Gate, asked when a successful trial deserves to become standard practice.

This volume asks what happens after the routine becomes ordinary.

Is the learner still using the mechanism that earned adoption—or has everyday life quietly turned it into something else?

Quick Read

The Drift Check compares the adopted routine with the routine actually being used now.
Drift is not the same as ordinary variation. One unusual lesson or one missed routine is not automatically drift.
Drift becomes meaningful when change is persistent, directional or large enough to alter the educational function.
A routine can drift in dose, timing, prompting, difficulty, task selection, feedback, ownership, scope or cost.
The key distinction is between the mechanism that must remain stable and the surface form that may adapt.
Good adaptation changes the surface because the learner or environment changed while protecting the educational job.
Bad drift changes the educational job without anyone consciously deciding to do so.
Every adopted routine should have a small tolerance band: enough flexibility for real life, but not so much that every deviation becomes normal.
The tutor should inspect direction, not only thresholds. A slow increase in prompting can matter before dependence becomes obvious.
Temporary exceptions need expiry conditions. Otherwise exceptions accumulate into a new operating system.
A useful Drift Check asks what was adopted, what is happening now, what changed, why it changed, and whether the change preserves the original function.
Sometimes the right response is restoration: return to the adopted route.
Sometimes the right response is deliberate re-baselining: the learner has grown and the old route should be updated.
Sometimes the right response is redesign: the original mechanism no longer fits the current problem.
Sometimes the right response is retirement: the learner no longer needs the routine.
Learning, studying, training, education and improvement drift in different ways and therefore require different sensors.
Class 0 to Class 6 tutor functions create different drift risks, from help creep to whole-system complexity.
Three-student tutorials are useful drift sensors because tutor attention moves and hidden dependence becomes visible.
A drift check should not become surveillance. It should be lightweight enough to preserve learning time.
The final question is not “Are we following the routine perfectly?” It is “Are we still protecting the function that made the routine worth adopting?”

1. What This Volume Owns

This volume owns a narrow tutor-side operating problem:

After a learner-specific routine has been adopted into ordinary tutoring, how does the tutor detect whether its lived form is quietly moving away from the mechanism that justified adoption?

The object can be a checking routine, retrieval opener, writing frame, study plan, prompt rule, mixed-practice sequence, paper-navigation method, correction process, parent-support boundary, homework-start routine, feedback loop or any other repeated tutoring structure.

The Drift Check is not interested in cosmetic perfection. It is interested in whether the adopted routine still performs the educational job it was commissioned to perform.

2. What This Volume Does Not Own

The general systems concept of drift belongs to broader eduKateSG owners such as How to Simplify Life | Drift Detection — Notice When a Good System Quietly Stops Being the System You Designed.

Implementation fidelity as a general educational problem belongs to How Implementation Fidelity Works | Preserve the Active Ingredients Without Turning Teaching Into a Script. eduKateSengkang’s Bolt Measurement Note 27 — You Cannot Judge an Intervention Before You Know What Was Actually Implemented owns the calibration problem of distinguishing intervention failure from implementation failure, especially at school-programme level.

How Instructional Variability Works owns variation in enacted instruction across teachers, classrooms and time.

The Tutor Handbook Vol No.0046 | The Stability Window owns how long a material route change should be held before the next serious judgement.

This volume stays narrower than all four. It owns the periodic learner-specific check that asks whether an adopted tutoring routine is still recognisably the same educational mechanism after ordinary use, adaptation and pressure have had time to change it.

3. Adoption Creates a Baseline

The Adoption Gate did more than say “keep this”.

It created a new reference point.

At adoption, the tutor should know what the routine is for, what version is being kept, how much support remains, what cost is acceptable, what boundary limits its use and what evidence should later make it shrink or disappear.

That record becomes the baseline against which drift can be detected.

Without a baseline, every current version can claim to be the intended version merely because it is the version currently happening.

If we cannot remember what the routine was meant to protect, we cannot tell whether adaptation improved it or quietly changed its purpose.

4. Mechanism and Surface Must Be Separated

A strong adopted routine has at least two layers.

Mechanism: the educational function that produced the benefit.
Surface: the visible form used to carry that function.

Beatrice’s visible surface might be two planning questions. The mechanism is idea–evidence control before drafting.

Alicia’s surface might be a three-item checklist. The mechanism is deliberate high-risk verification before a multi-step answer leaves working memory.

Faith’s surface might be a four-step help gate. The mechanism is learner initiation before precise escalation.

The surface is allowed to evolve. The mechanism should not disappear accidentally.

5. Variation Is Not Automatically Drift

Real tutoring is adaptive.

A tutor may shorten a checklist on an easy task, extend wait time on a difficult task, postpone timed work when a new concept is being learned, or change examples because the first representation did not connect.

That is not necessarily drift.

Drift appears when the variation begins to change the route persistently without a deliberate decision.

One prompt during an unusually difficult task is variation.

The tutor gradually giving the first step on every difficult task is prompt drift.

One extra timed set before a school test is variation.

Timed work quietly replacing untimed learning across most lessons is timing drift.

6. Deliberate Adaptation Is Different From Accidental Drift

Deliberate adaptation has a reason that can be stated.

“We are shortening the planning frame because Beatrice can now reconstruct the first two steps independently.”

“We are increasing mixed practice because execution is stable and selection is now the active bottleneck.”

“We are temporarily restoring one prompt because the school has introduced a new task form and the learner cannot yet parse it reliably.”

Accidental drift usually has no clean sentence.

It sounds like:

We just sort of started doing it this way.

That sentence should wake the tutor up.

7. The Drift Check Equation

Adopted Mechanism + Allowed Adaptation + Current Enacted Routine + Current Receipts + Learner Ownership = Drift Classification.

The tutor compares five things.

What mechanism crossed the Adoption Gate?
What kinds of variation were expected or allowed?
What is actually happening now?
Are the original learning receipts still appearing?
Has ownership moved toward or away from the learner?

The comparison produces one of four broad conclusions: stable, usefully adapted, drifting, or obsolete.

8. Stable Does Not Mean Identical

A stable route can look different from week to week.

Alicia may use the full check on a long algebraic manipulation and a mental scan on a short substitution.

Beatrice may plan with written questions for a difficult essay and mentally for a familiar paragraph task.

Faith may use her full help gate for unfamiliar work and ask immediately when the instruction itself is ambiguous.

The form varies.

The function remains recognisable.

9. Every Routine Needs a Tolerance Band

If the tutor reacts to every small deviation, the learner’s system becomes exhausting.

One missed retrieval session is not necessarily drift.

One rushed evening is not necessarily study-system failure.

One prompt during a difficult question is not necessarily restored dependence.

A tolerance band says ordinary variation is allowed until persistence, direction or magnitude becomes educationally meaningful.

Variation lives inside the tolerance band. Drift is persistent movement that begins changing the educational job.

10. Direction Can Matter Before a Threshold Breaks

Suppose Faith’s adopted routine is learner-first initiation.

Week one: tutor supplies the first move in one of ten unfamiliar questions.

Week two: two of ten.

Week three: four of ten.

Faith may still look broadly independent.

The direction is already informative.

Good tutors do not wait for complete dependence before asking why the support slope is moving.

11. Drift Type One: Dose Drift

A routine can keep the same name while its dose expands.

Ten minutes of retrieval becomes twenty-five.

One checking cycle becomes checking after every question.

One parent review each fortnight becomes nightly monitoring.

One full paper every two weeks becomes three papers a week.

The mechanism may still be sensible.

The dose may no longer be.

12. Drift Type Two: Prompt Drift

Prompt drift is especially quiet because helpful tutors naturally respond when a learner hesitates.

A short cue becomes a leading question.

A leading question becomes the first step.

The first step becomes an explanation before the learner has attempted anything.

The worksheet still gets finished.

The learner’s control boundary has moved.

13. Drift Type Three: Task-Selection Drift

A routine may be adopted for high-value representative tasks and then drift toward whatever is easiest to administer.

Mixed practice becomes chapter worksheets again because they are quicker to prepare.

Fresh comprehension passages become familiar school corrections because they are already available.

Independent writing becomes heavily scaffolded model imitation because marking is faster.

The activity continues.

The evidence condition weakens.

14. Drift Type Four: Difficulty Drift

Difficulty can drift upward or downward.

Downward drift happens when the tutor keeps selecting easier work because success feels productive.

Upward drift happens when every success is answered with harder work until the learner is constantly operating at the edge of overload.

The adopted mechanism may have been “use representative challenge”.

Drift turns it into “keep success comfortable” or “keep challenge escalating”.

Neither is the original job.

15. Drift Type Five: Timing Drift

Time pressure is a powerful training variable.

It is also easy to over-apply.

One representative timed block becomes the default condition for practice.

Eventually the learner begins first exposure, deep repair and unfamiliar reasoning under a clock that was originally intended only for performance conversion.

The timer did not merely become more frequent.

It crossed into a different learning layer.

16. Drift Type Six: Feedback Drift

A feedback routine begins with one clear improvement instruction followed by a reattempt.

Over time, the comments become longer. The tutor marks more details. The learner reads more and changes less.

The original mechanism was feedback → action → new performance.

The drifted mechanism is tutor commentary → learner receipt.

More feedback has produced less feedback loop.

17. Drift Type Seven: Ownership Drift

Ownership drift is one of the most important Handbook signals.

A learner-owned plan becomes tutor-updated.

A tutor-updated plan becomes parent-enforced.

A learner-owned correction becomes tutor-marked and parent-checked.

The visible system may become more reliable in the short term.

The locus of control moves away from the learner.

18. Drift Type Eight: Measurement Drift

A routine may originally be evaluated on independent fresh performance.

Later, the tutor starts using easier evidence because it is convenient.

Homework replaces fresh probes. Familiar retests replace changed questions. Heavily prompted work is counted alongside unsupported work. Completion becomes a proxy for capability.

The learner may not have changed.

The measuring condition has.

This volume does not own measurement theory. It simply asks the tutor to notice when the evidence used to justify an adopted route no longer resembles the evidence that originally proved the route useful.

19. Drift Type Nine: Scope Drift

A narrow tutoring routine can gradually annex neighbouring work.

A study-planning audit becomes general homework management.

A Mathematics repair becomes daily schoolwork supervision.

A writing review becomes rewriting every school assignment.

A performance routine becomes a standing judgement about the learner’s confidence, motivation and life organisation.

The Scope Boundary should be reopened when a routine quietly expands beyond the job it was adopted to do.

20. Drift Type Ten: Cost Drift

An adopted routine may keep producing the target benefit while becoming progressively more expensive.

A fifteen-minute weekly review becomes forty minutes.

A retrieval routine expands from ten items to thirty.

One additional full paper becomes a weekend expectation.

The routine still “works”.

The opportunity cost has changed.

The Intervention Cost should be reread whenever an adopted routine becomes materially larger.

21. Drift Type Eleven: Exception Accumulation

Many systems drift through reasonable exceptions.

“Just this week, I’ll choose the revision tasks for you.”

“Because the test is near, we’ll skip the independent first attempt.”

“This composition is important, so I’ll give you the outline.”

Each exception can be sensible.

The risk appears when the exception has no expiry condition and keeps returning.

A temporary exception without a return condition is a candidate new default.

22. Drift Type Twelve: Complexity Drift

Good routines can accumulate accessories.

The study plan gets a tracker.

The tracker gets a colour code.

The colour code gets a weekly audit.

The audit gets a parent signature.

The learner is now running an administrative system around studying.

The original mechanism may have been simple priority control.

Complexity drift makes the support harder to maintain than the learning it was designed to protect.

23. Drift Can Be Positive

Not every departure from the original adopted form should be reversed.

Alicia may stop looking at the printed checklist because the three checks have become automatic.

Beatrice may stop writing planning questions because she can mentally generate the relationship quickly.

Faith may no longer need the full help gate because she has learned to distinguish productive struggle from an unclear instruction.

The system changed.

The learner became more capable.

The correct move is not restoration.

It is deliberate re-baselining.

24. Re-Baselining Must Be Deliberate

A baseline should not become nostalgia.

The learner changes.

The curriculum changes.

The examination season changes.

A routine that was ideal during repair may be wasteful during maintenance. A routine that was sufficient in Primary 6 may need a new version in Secondary 1. A performance strategy that worked on short sets may need adaptation for full papers.

The Drift Check therefore asks two questions before restoring an old state:

Did the routine drift away from a still-useful design?
Or did the learner outgrow the old design?

25. Learning Drift

Within the eduKate architecture, learning concerns changes in understanding, retrieval, representation, reasoning and usable knowledge.

Learning routines drift when the support starts carrying the operation the learner was meant to learn.

An explanation routine drifts into answer-giving.

A worked example routine drifts into permanent imitation.

A retrieval routine drifts into rereading immediately before every check.

A representation routine drifts into “always draw this diagram” rather than “choose a representation that exposes the relation”.

The learning Drift Check asks:

Is the learner still generating more of the capability—or is the routine increasingly generating it for them?

26. Studying Drift

Studying concerns planning, time, resources, retrieval, review and self-management outside the immediate teaching act.

Study routines are especially vulnerable to administrative drift.

A simple priority list becomes a full timetable.

The timetable becomes a tracker.

The tracker becomes a parent-reporting system.

The learner spends increasing effort maintaining evidence that studying happened.

The studying Drift Check asks:

Is this routine still helping the learner decide and act—or has managing the routine become another task?

27. Training Drift

Training concerns making available capability more reliable.

Training drift often appears as volume creep.

Ten useful repetitions become forty familiar repetitions.

One moderate timed block becomes constant performance pressure.

One narrow drill becomes the entire revision strategy.

The training Drift Check asks:

Is repetition still widening reliability, or are we merely accumulating familiar performance?

28. Education Drift

Education places the learner inside real curricula, school stages, assessments, deadlines and institutional conditions.

An adopted tutoring route can drift away from that real environment.

The tutor keeps teaching a preferred sequence while school moves elsewhere.

The learner practises one answer form while the school’s current task demands another.

A study system designed for a light term remains unchanged when examination load rises.

The education Drift Check asks:

Does the adopted route still fit the educational environment the learner actually has to navigate now?

29. Improvement Drift

Improvement is the movement from observation through diagnosis, prioritisation, repair, practice, connection, performance and review.

Improvement drift happens when the loop gets stuck in the stage that once helped.

Diagnosis never ends.

Repair never reaches transfer.

Practice never reaches performance.

Performance coaching never returns to ordinary learning.

The improvement Drift Check asks:

Is the learner still moving through the improvement loop, or has one useful stage become permanent residence?

30. Class 0 · Homework Helper Drift

In the Tutor Classification Model by eduKateSG, Class 0 works nearest to task completion and routine.

The main drift risk is help creep.

The tutor begins by clarifying one instruction.

Then starts opening the school platform.

Then identifies the deadlines.

Then chooses which assignment to begin.

The learner’s completion may improve while their task-management function becomes more external.

Class 0 Drift Check: What adult action has quietly become part of ordinary task completion that the learner was previously expected to carry?

31. Class 1 · Explainer Drift

The Explainer’s drift risk is explanation expansion.

A concise representation that once unlocked the concept grows into a full tutor monologue because the tutor wants to ensure complete clarity.

The learner gets more explanation and fewer opportunities to reconstruct.

Class 1 Drift Check: After the explanation, is the learner doing more reconstruction over time—or is the tutor doing more explaining?

32. Class 2 · Drill Builder Drift

The Drill Builder’s drift risk is volume without a changing evidence condition.

The learner gets faster on the same surface.

The tutor responds with more of the same.

Eventually the drill is no longer testing the bottleneck that justified the drill.

Class 2 Drift Check: What changed in the learner that should now cause the practice condition to change?

33. Class 3 · Diagnostic Tutor Drift

The Diagnostic Tutor’s drift risk is confirmation.

An early explanation becomes familiar.

New evidence gets interpreted through it.

Probes begin confirming the existing diagnosis rather than trying to separate it from plausible alternatives.

Class 3 Drift Check: What evidence would make the current diagnosis weaker, and have we still allowed that evidence a chance to appear?

34. Class 4 · Route Designer Drift

The Route Designer’s drift risk is priority inertia.

A route that was correct last month keeps receiving capacity because the plan already exists.

Meanwhile the learner’s school demands, prerequisites or bottlenecks change.

Class 4 Drift Check: If we built the route today from the current evidence, would the next three priorities still be the same?

35. Class 5 · Performance Coach Drift

The Performance Coach’s drift risk is exam conditions spreading into all learning.

Everything becomes timed.

Everything becomes marks.

Every error is treated as lost performance rather than information about a learning mechanism.

Class 5 Drift Check: Are we still coaching performance where performance conditions matter, or has performance pressure colonised learning?

36. Class 6 · Learning Architect Drift

The Learning Architect’s drift risk is architecture growth.

Every problem gets a process.

Every process gets a record.

Every record gets a review.

The learner’s system becomes sophisticated and difficult to live in.

Class 6 Drift Check: Which parts of the architecture can now disappear because the learner has become stronger?

37. Repair Mode Drift

eduKateSG’s Three Modes of Tuition separate Repair, Alignment and Frontier work.

Repair drifts when temporary narrow rebuilding becomes the learner’s permanent mode.

The learner keeps revisiting foundations even after the downstream receipt has stabilised.

The Drift Check asks whether repair can now collapse into maintenance, mixed use or ordinary curriculum participation.

38. Alignment Mode Drift

Alignment drifts when staying connected to school becomes following school reactively at all costs.

Every homework deadline wins.

Every upcoming test displaces foundational work.

The tuition route becomes a second pair of hands for current school demand rather than a system that protects present participation and long-term capability together.

39. Frontier Mode Drift

Frontier work drifts when “stretch” becomes “more difficult because more difficult looks advanced”.

The learner accumulates exotic problems without deeper transfer, explanation or independent problem selection.

The Drift Check asks whether frontier work is expanding capability or merely decorating the programme with difficulty.

40. Three-Student Tutorials Are Natural Drift Sensors

In a three-student tutorial, the tutor’s attention moves.

That creates repeated moments when the adopted routine must operate without immediate tutor activation.

Does Alicia still run the check when the tutor is with Beatrice?

Does Faith still begin the unfamiliar task when the tutor cannot respond instantly?

Does Denise manage the timing checkpoint without hearing it announced?

Does Emily update her plan while the tutor is checking another learner’s paper?

These are low-cost implementation receipts.

41. Group Routines Can Drift Into Group Uniformity

A routine that helps one learner is easy to spread to the whole table.

Everybody receives the same checklist.

Everybody follows the same timing rule.

Everybody uses the same paragraph frame.

The class becomes easier to manage.

Individual adoption evidence disappears.

Shared teaching is valuable. Shared routines should still have learner-specific reasons.

42. One-to-One Tutoring Can Hide Drift

One-to-one tutoring is so responsive that the tutor can continuously compensate for a drifting routine.

The learner forgets the check; the tutor reminds them.

The learner does not start; the tutor reframes the task.

The study plan becomes unrealistic; the tutor silently fixes it.

Output remains strong because the adult absorbs the drift.

One-to-one Drift Checks therefore need deliberate tutor absence: silent first attempts, learner-led planning, delayed feedback and work completed between sessions.

43. Alicia · The Checklist That Became a Glance

Alicia adopted a sign–unit–copied-value scan for high-risk multi-step Mathematics.

Six weeks later the tutor notices that the routine takes two seconds.

That could mean fluency.

Or it could mean ritual.

The tutor does not tell Alicia to slow down automatically.

Instead, one representative set is used to inspect what she actually checks. Alicia catches units reliably, but sign checking has become a general glance at the final answer rather than a scan of risky transformations.

The mechanism has partially drifted.

Repair: restore sign checking only at the specific transformation points that historically produced error. The routine becomes more precise, not simply longer.

44. Beatrice · The Planning Rule That Moved After the Draft

Beatrice still writes her two planning questions.

The tutor initially assumes the adopted routine is intact.

Then a fresh writing sample reveals that Beatrice drafts the paragraph first and fills in the questions afterward because the school deadline feels urgent.

The artifact is present.

The causal order is reversed.

The original mechanism was planning → drafting.

The drifted mechanism is drafting → retrospective paperwork.

Repair: remove the requirement to write the planning questions. Ask Beatrice to state them orally before drafting. The tutor protects the mechanism and removes the bureaucratic incentive to fake the artifact.

45. Ciara · The Science Rule That Became a Template

Ciara adopted a strong principle: when a Science condition changes, rebuild the causal mechanism before reusing model wording.

Later, she starts drawing the same arrow diagram for every explanation.

The diagram once supported reasoning.

Now the diagram has become a template independent of whether the relation actually needs it.

The Drift Check reveals strategy rigidity.

Repair: ask Ciara to choose whether the task needs a causal chain, comparison table, labelled diagram or direct verbal relation before she starts. The adopted mechanism becomes representation choice, not one favourite representation.

46. Denise · The Timed Block That Ate the Lesson

Denise’s adopted routine was one short weekly timed Additional Mathematics block.

It worked.

As prelims approach, the tutor adds another block. Then a full section. Then a paper.

Denise’s completion improves, but new topics are now introduced under time pressure and conceptual errors are being treated as performance errors.

The original timing routine has crossed layers.

Repair: preserve timed work for already-learned representative performance. Return new learning and weak-link repair to untimed conditions. Examination proximity changes the budget; it does not erase the distinction between learning and performance.

47. Emily · The Rolling Plan That Became a Full Calendar

Emily adopted a rolling three-day plan because fixed seven-day timetables broke under school variability.

After several good weeks, she starts adding every possible task into the rolling plan “so nothing gets forgotten”.

The plan becomes a three-day list of seven-day obligations.

The original mechanism was bounded prioritisation with room for adaptation.

The drifted mechanism is exhaustive task capture.

Repair: separate the complete task inventory from the active three-day plan. The inventory remembers. The plan decides.

48. Faith · The Help Gate That Became a Permission Rule

Faith’s help gate was designed to improve initiation and precise help-seeking.

Over time, “try one plausible route first” becomes “you must not ask until you have struggled for five minutes”.

The learner starts persisting when the instruction itself is unclear.

The route now trains endurance instead of help calibration.

Repair: restore the original distinction. If the task is understood, attempt one route. If the task condition is genuinely unclear, ask immediately and specifically.

Independence is not delayed rescue.

It is better judgement about when rescue is needed.

49. The Sixty-Second Drift Check

What was this routine originally for?
What is actually happening now?
What changed in dose, prompting, timing, task type or ownership?
Are the original learning receipts still appearing?
Is the change deliberate adaptation, ordinary variation, harmful drift or learner growth?
Restore, re-baseline, redesign or retire?

That is enough for many routines.

50. The Full Drift Card

Routine: What adopted practice are we checking?
Original mechanism: What educational job earned adoption?
Original version: What did the adopted route look like?
Allowed adaptation: What changes were expected to be harmless or useful?
Current enacted version: What actually happens now?
Dose: More, less or the same?
Support: More, less or different?
Task condition: Same evidence quality or easier/harder/familiar?
Ownership: More learner-owned or more adult-owned?
Cost: Has recurring time or attention expanded?
Receipt: Is the original benefit still present?
New benefit: Has useful adaptation created something better?
New harm: What collateral effect has appeared?
Classification: Stable, adapted, drifting or obsolete?
Decision: Restore, re-baseline, redesign or retire?
Next check: What evidence will show whether the decision worked?

51. Do Not Audit Everything Every Week

A Drift Check becomes harmful if every routine becomes another monitoring project.

Use it where drift matters.

high-cost routines;
routines that control substantial tutor support;
routines that can create dependence;
routines near examinations;
routines that have accumulated exceptions;
routines whose visible artifact remains but whose educational receipt seems weaker;
routines that used to feel simple and now feel heavy.

Monitoring should earn the time it consumes.

52. Use Triggered Checks, Not Constant Surveillance

Useful drift triggers include:

a sudden rise in prompting;
a routine taking much longer than before;
a learner saying “I still do it” while the evidence looks different;
a new school stage or examination season;
a parent becoming increasingly involved in a routine that was learner-owned;
a support being used outside the context for which it was adopted;
the original learning receipt weakening;
the learner succeeding despite not using the visible scaffold;
a temporary exception repeating for several cycles.

Triggers make the Drift Check event-driven rather than bureaucratic.

53. Restore When the Mechanism Is Still Right

Restoration is appropriate when the adopted mechanism still fits the learner and the environment, but ordinary shortcuts or pressure have weakened its enactment.

Beatrice still needs planning before drafting; she simply moved the planning artifact after the draft.

Restore the causal order.

Denise still benefits from representative timed work; the clock simply spread into learning conditions where it does not belong.

Restore the boundary.

54. Re-Baseline When the Learner Has Grown

Re-baselining is appropriate when the old routine has become unnecessarily heavy because the learner now carries more capability.

The printed checklist becomes a mental scan.

The tutor-led study review becomes a learner-prepared audit.

The full help gate becomes one internal question.

The original routine did not fail.

It succeeded well enough that the next stable state can be lighter.

55. Redesign When the Environment or Bottleneck Changed

Redesign is appropriate when the learner still needs support but the original mechanism no longer matches the controlling problem.

A drill that once improved fluency now reveals method-selection weakness.

A weekly timetable that once solved task-starting now fails because the new school stage creates variable workload.

A full-paper routine that once improved endurance now produces flat returns because the active bottleneck is one narrow algebraic family.

Do not restore a routine merely because it was once successful.

56. Retire When the Routine No Longer Earns Its Cost

Retirement is appropriate when the learner can carry the target capability without the routine, or when ordinary curriculum use now maintains it cheaply.

A stable capability does not need a permanent museum of the scaffold that built it.

The earlier Tutor Handbook Vol No.0033 | The Retirement owns the decision to end special monitoring of a known fragility. The same restraint applies here: routines should not survive merely because they have become familiar.

57. Parent Communication: “It Has Changed” Is Not Necessarily Bad News

A useful parent explanation can sound like this:

We adopted this routine because it helped your child start unfamiliar tasks without waiting for me. The routine is still present, but I have noticed that I am supplying more of the first step again. I am checking whether that is a temporary response to harder school work or a real return of support dependence. I will restore the learner-first condition on representative tasks and review the next few attempts before changing the wider plan.

The parent hears evidence, uncertainty and next action.

There is no need to turn drift into drama.

58. Student Communication: Teach the Learner to Detect Their Own Drift

Eventually the learner should be able to notice when a useful routine changes function.

Ask:

What was this routine supposed to help you do?
What do you actually do now?
Which part has become automatic?
Which part have you stopped doing?
Did you stop because you no longer need it or because it became inconvenient?
What happens to performance when the routine disappears?
Could the routine become shorter without losing its job?

This is metacognitive maintenance.

The learner is not merely following a system.

They are learning to inspect the system they follow.

59. Near an Examination, Drift Speeds Up

Examination proximity creates legitimate pressure for more representative performance work.

It also creates drift risk.

More papers.

More timing.

More parent involvement.

Less sleep.

Less space for foundational repair.

The Drift Check near an examination asks which changes are deliberate seasonal reallocation and which are panic becoming infrastructure.

60. After the Examination, Recheck the Temporary Changes

Performance season often creates temporary exceptions.

Those exceptions need a return path.

If a parent resumed closer monitoring, does that monitoring end?

If full papers expanded, does the programme return to learning and repair afterward?

If bedtime shifted, does the study plan re-expand recovery?

The end of the examination is a natural drift reset point.

61. Transitions Require Fresh Drift Tolerances

Primary to Secondary, Secondary 2 to Secondary 3, new subject combinations and new examination years all change the environment.

An adopted routine may need temporary expansion while the learner adapts.

The tutor should not treat every increase in support as regression.

The question is whether the expansion has an educational reason, a review condition and a path back toward learner ownership.

62. Technology and AI Create Fast Drift

Digital tools can change a learner’s routine rapidly because the tool can absorb operations invisibly.

A spellchecker begins as final verification and becomes first-draft dependence.

An AI explanation begins as clarification and becomes the default before the learner attempts the text.

A solution generator begins as post-attempt comparison and becomes the route selector.

The tool itself is not the problem.

The Drift Check asks which learner operation the tool was originally allowed to carry and whether that boundary has moved.

63. Research Foundation: Implementation Must Be Compared With What Was Intended

The U.S. Institute of Education Sciences’ June 2023 guide Conducting Implementation Research in Impact Studies of Education Interventions recommends explicitly examining whether and how an intervention was implemented, the context in which implementation occurred, and the contrast between what was planned and what people actually received.

The Tutor Handbook is operating at the scale of one tutor and one learner rather than a formal impact study.

The transferable principle is modest and powerful:

Before judging the result of a routine, know what routine was actually enacted.

64. Research Foundation: Identify Core Elements, Then Allow Adjustment

The Australian Education Research Organisation’s Developing a School Implementation Plan, published in March 2025 and updated in April 2026, asks implementation teams to identify the core elements of a practice, monitor implementation outcomes, schedule activities and adjust when needed.

Again, a tutoring routine is much smaller than school-wide implementation.

But the same design idea helps: know what must remain intact, know what can adapt and use evidence to decide whether the current form still protects the intended function.

65. Research Foundation: Implementation Should Be Structured but Flexible

The Education Endowment Foundation’s third-edition A School’s Guide to Implementation, published in April 2024, organises implementation into Explore, Prepare, Deliver and Sustain while explicitly describing the process as structured but flexible.

The Drift Check lives inside the same tension.

Too much rigidity prevents adaptation.

Too little structure makes the original active ingredient disappear.

The tutoring challenge is to protect function without turning teaching into a script.

66. Research Foundation: Monitor Progress So Evidence Can Change Teaching

AERO’s Monitor Progress guidance emphasises checking what learners know and can do and adjusting instruction, guidance or feedback when the evidence requires it.

A Drift Check should therefore not become fidelity theatre.

The purpose is not to prove that everybody followed the old routine.

The purpose is to decide whether the learner still needs that routine, needs a restored version, needs a better version or no longer needs it at all.

67. Research Foundation: Reflection Is Part of Sustainment

EEF’s implementation guidance treats reflection, monitoring and adaptation as part of effective implementation rather than as signs that the original design was weak.

That matters in tutoring because a route that cannot be inspected is not truly stable.

Stable practice is not frozen practice.

It is practice whose changes remain accountable to the reason it exists.

68. Drift and the Decision Record

The Decision Record makes the Drift Check possible.

If the tutor recorded why the route was adopted, what receipt was expected, what support remained and what would later justify change, drift can be detected against something concrete.

If no record exists, reconstruct the smallest useful baseline now.

Do not let missing documentation become an excuse for never inspecting the route.

69. Drift and the Learning Claim

The Learning Claim prevents drift from being diagnosed through vague impressions.

“The routine is not working anymore” is too broad.

“Independent starts have fallen from most unfamiliar tasks to about half while first-step tutor prompts have increased” is a usable claim.

Drift detection becomes better when the changed state can be described rather than merely felt.

70. Drift and the Learning Budget

Every drifted routine changes allocation.

71. Drift and the Stability Window

The Stability Window and Drift Check can appear contradictory.

One says hold the route steady.

The other says inspect whether the route changed.

They are complementary.

The Stability Window prevents overreaction to noise.

The Drift Check prevents “hold steady” from becoming blindness to the fact that the enacted route is no longer steady at all.

Hold the intended mechanism steady long enough to learn from it, while checking that ordinary adaptation has not quietly changed the mechanism being held.

72. Drift and the Adoption Gate

The Adoption Gate asks whether a successful trial deserves a stable place in ordinary practice.

The Drift Check asks whether ordinary practice is still recognisably what crossed that gate.

Together they create a complete loop:

Trial → Adopt → Enact → Check Drift → Restore / Re-Baseline / Redesign / Retire → Continue.

A routine is no longer a one-time decision.

It becomes a living part of the learner’s system.

73. What the Drift Check Is Not

It is not a demand for perfect compliance with the original routine.
It is not a reason to punish adaptation.
It is not a weekly audit of every tutoring behaviour.
It is not a teacher-observation rubric.
It is not a diagnosis of the learner from one bad week.
It is not proof that any deviation is harmful.
It is not a rule that the old baseline is always better than the new state.
It is not a reason to preserve obsolete scaffolds.
It is not the same as implementation fidelity at school-programme scale.
It is not the same as instructional variability across teachers.
It is not a commercial reason to keep tuition running.
It is a lightweight check that protects the educational function of a routine after ordinary life has had time to change it.

74. The Tutor’s Drift Check Checklist

What did we originally adopt?
What educational mechanism made it worth adopting?
What surface features were always meant to be flexible?
What is the learner actually doing now?
Has the dose changed?
Has the amount of prompting changed?
Has the task family become easier, more familiar or less representative?
Has time pressure spread into a different learning layer?
Has feedback grown while learner action shrank?
Has ownership moved toward the learner or back toward adults?
Has the routine annexed work outside its original scope?
Has the recurring time or attention cost increased?
Are exceptions becoming defaults?
Has the system become unnecessarily complex?
Are the original receipts still present?
Is there a new benefit the old baseline did not anticipate?
Is the learner succeeding without part of the scaffold?
Is this ordinary variation, deliberate adaptation, harmful drift or growth?
Should we restore, re-baseline, redesign or retire?
What next evidence will tell us whether that decision was correct?

75. The Parent’s Drift Check Checklist

What routine did tuition originally introduce?
What was it supposed to improve?
Is my child using it independently now?
Has the routine become bigger or more time-consuming?
Has parent involvement increased without a new decision?
Is the routine still improving the original target?
Has a temporary exam-period exception become normal?
Does the tutor explain whether changes are deliberate?
Can any part of the support now shrink?
If the routine has changed, is the new version actually better for the learner?

76. The Learner’s Drift Check Checklist

What was this routine meant to help me do?
What do I actually do now?
What part has become automatic?
What part do I skip?
Did I skip it because I no longer need it or because I am rushing?
Do I wait for more help now than when the routine started?
Does the routine still help when the task looks different?
Is the routine taking more time than it used to?
Can I use a shorter version?
Do I know when not to use it?
If the routine disappeared tomorrow, what useful capability would remain?

77. The Ethical Standard

Tutoring is full of useful temporary structures.

The danger begins when temporary structures survive without being re-examined, when helpful adaptations quietly change ownership, or when an old success becomes authority for a present routine nobody would consciously design today.

A tutor should not demand rigid obedience to yesterday’s method.

A tutor should not ignore the purpose of yesterday’s method either.

The professional obligation is to know what function mattered, allow the surface to evolve, detect when the lived routine changes the function, and return as much control as the learner can now carry.

Do not preserve the routine because it is familiar. Preserve the function because it is useful. When the learner or environment changes, update the route deliberately rather than letting the route change itself.

Evidence and Connected Reading

Final Compression

The routine was adopted for a reason.

Remember the reason.

Separate mechanism from surface.

Allow ordinary variation.

Allow deliberate adaptation.

Watch the direction.

Notice dose creep.

Notice prompt creep.

Notice when task quality changes.

Notice when timing spreads.

Notice when feedback stops producing action.

Notice when ownership moves backward.

Notice when exceptions become defaults.

Notice when the routine costs more than the version that was adopted.

Do not restore an obsolete baseline merely because it is documented.

Ask whether the learner has grown.

Restore when the mechanism is still right.

Re-baseline when capability has grown.

Redesign when the bottleneck or environment changed.

Retire when the routine no longer earns its place.

Then write the new state clearly enough that tomorrow’s drift can be detected too.

The Drift Check protects a successful tutoring routine from two opposite errors: rigidly preserving a form the learner has outgrown, and casually allowing everyday shortcuts to erase the mechanism that made the routine successful. Keep the function visible, let the form evolve deliberately, and never let “this is how we do it now” become the only reason a learning system continues.

That is the Drift Check.

That is Tutor Handbook Volume 0054.

Small Group Tutorials