Zone of Proximal Development: Vygotsky's ZPD, With Examples

Go from storyboard to an interactive course in minutes

Start Free

Picture two new receptionists at a vet clinic, a week into the job. A caller says her terrier has eaten most of a block of dark chocolate and seems fine, and both of them offer her an appointment later that day. On paper they are at the same level, because both got it wrong alone. Then a senior receptionist helps each of them through a similar call. One needs only a nudge to reread what the caller said. The other needs to see the senior receptionist's full call note before the urgency makes sense.

The gap between what each receptionist could do alone and what they could do with help illustrates a later workplace adaptation of Lev Vygotsky's zone of proximal development. I'm an instructional designer at Coursebox, where we make instructional design software that turns a brief into a course outline, so I'll mention our product where I tested it. Here I explain it in Vygotsky's own words, correct the ideas that have been attached to it since, and show how to design training around it, with three real studies, a worked example from a vet clinic front desk and two free templates.

What is the zone of proximal development?

The zone of proximal development (ZPD) is Lev Vygotsky's name for the gap between what a learner can do alone and what they can do with help from an adult or a more capable peer, a gap that shows which abilities are still maturing.

View the full-size diagram

Vygotsky's own definition is in Mind in Society, the 1978 English collection of his writings. The zone is "the distance between the actual developmental level as determined by independent problem solving and the level of potential development as determined through problem solving under adult guidance or in collaboration with more capable peers" (p. 86).

The idea starts as a complaint about testing. "We must determine at least two developmental levels," Vygotsky wrote (p. 85), because the mental tests of his day measured only what a child could do alone. His example was two children, both "ten years old chronologically and eight years old in terms of mental development" (p. 85). Then, he went on, suppose the testers help them: they might run through a demonstration, start the solution for the child to finish or ask leading questions (p. 86). With that help, the first child could deal with problems up to a twelve-year-old's level and the second only up to a nine-year-old's.

Alone, the two children looked the same. With help, one was four years ahead of the test score and the other one year. "This difference between twelve and eight, or between nine and eight, is what we call the zone of proximal development," Vygotsky wrote (p. 86). A test that records only the first number would put both children in the same class and teach them the same way.

What the zone shows is development that is under way. Vygotsky said it "defines those functions that have not yet matured but are in the process of maturation", the "buds" or "flowers" of development rather than the "fruits" (p. 86). That makes it forward-looking: "what a child can do with assistance today she will be able to do by herself tomorrow" (p. 87). The zone also has an upper edge. In Thinking and Speech he wrote that "Ultimately, problems become too difficult to resolve even in collaboration" (p. 209), and in Mind in Society that "a person can imitate only that which is within her developmental level" (p. 88). Help works inside the zone, not at any distance from what someone can already do.

Vygotsky (1896 to 1934) was a Soviet psychologist who described the zone in lectures and papers in the early 1930s, before he died of tuberculosis in 1934. He was writing about how children's thinking develops, at school and in play, where he argued that "play creates a zone of proximal development of the child" (Mind in Society, p. 102). English readers mostly know the idea from Mind in Society, which its editors called "not a literal translation of Vygotsky but rather our edited translation of Vygotsky". Applying the zone to adults learning job skills is a later extension of his idea. It is useful at work, as long as you treat it as an extension rather than something Vygotsky claimed.

Here are the parts of the idea at a glance, with the question each one asks of your training.

The clearest real example I found of training that missed people's zone comes from a health record rollout in Norway.

Case study: Learning a new health record system in a Norwegian nursing home

This was a study of an existing rollout through the lens of the ZPD, not a programme designed on it. In early 2023, a municipal nursing home in Central Norway moved to a new regional electronic health record built on Epic. Staff had e-learning and one eight-hour classroom day in a sports hall about three months before go-live, and staff shortages left little time to practise.

• What the researchers did: Wiggen and colleagues interviewed 12 people (10 nurses, a department manager and an implementation coordinator) and spent 6 hours observing two nurses who were superusers. They used the ZPD to analyse why the official training and the help that grew up on the ward worked so differently.
• Evidence and result: In their 2026 paper in the Scandinavian Journal of Primary Health Care, the training "appeared poorly suited to the actual capabilities of many employees". The authors concluded that it "was outside the development zone of many employees" and was "conducted far from the work context". What worked was colleagues helping each other on the ward, shoulder to shoulder. As one nurse put it, "It can take one minute, and it can take five." Superusers who had missed their own training were less able to help.
• Limitation: It is a qualitative study of 12 people at one site, interviewed soon after go-live, with no test scores, error rates or system data. The authors also warn that informal help can spread workarounds.
• Lesson: In this rollout, staff reported that the distant, feature-heavy training did not match their needs. For a similar rollout, pilot training close to go-live in small steps tied to real tasks, find out what each group can already do, and give superusers time to help colleagues when they get stuck.

Vygotsky's books on the ZPD: which to read

The zone takes up only a few pages of Vygotsky's work, and three texts cover it.

• Mind in Society (1978). Read chapter 6 first. "Interaction between Learning and Development" holds the definition, the two children and the argument about imitation in eight pages (pp. 84 to 91). It is an edited compilation, so you are reading Vygotsky through four editors.
• Thinking and Speech (1934), in Norris Minick's translation in volume 1 of the Collected Works (1987). Read chapter 6 for the fuller version. It includes the gardener who judges an orchard only by "the matured or harvested fruits" (p. 208). An abridged 1962 translation, Thought and Language, mentions the zone on only two pages.
• Seth Chaiklin, "The zone of proximal development in Vygotsky's analysis of learning and instruction" (2003). Read it to avoid the common misreadings. It is the clearest short account of what Vygotsky meant and what later writers added.

What the ZPD is not: four common mix-ups

Much of what is taught as Vygotsky's ZPD was added later. Four additions cause the most trouble, and each one changes what you would design.

"More knowledgeable other" is a later phrase

Vygotsky wrote about "adult guidance" and "more capable peers" (p. 86). The phrase "more knowledgeable other" does not appear anywhere in Mind in Society. It spread in American education writing in the early 1990s, and by 1995 at least one paper was printing it inside the p. 86 definition, as if Vygotsky had written it. The short form MKO seems to be later still. I could not find who coined the phrase, and the label is not the useful part anyway. Vygotsky cared about what a learner can do with help, not about the status of the helper. Chaiklin lists the belief that learning depends on "interventions by a more competent other" as one of three common misreadings of the zone, which he calls the "assistance assumption".

Scaffolding comes from Wood, Bruner and Ross

Scaffolding is not in Mind in Society either. The metaphor comes from David Wood, Jerome Bruner and Gail Ross's 1976 paper on tutoring, in which a tutor helped 30 children aged 3 to 5 build a pyramid from 21 blocks. They described scaffolding as the adult "controlling" the parts of a task "that are initially beyond the learner's capacity" (p. 90), and named six functions of the tutor (p. 98):

• Recruitment: getting the learner interested in the task and its requirements.
• Reduction in degrees of freedom: simplifying the task so the learner can handle it.
• Direction maintenance: keeping the learner working towards the goal.
• Marking critical features: pointing out the features that count.
• Frustration control: keeping the struggle manageable.
• Demonstration: showing a solution for the learner to imitate.

Their paper does not cite Vygotsky. Courtney Cazden linked scaffolding to the ZPD in 1979, and the two have travelled together since. Shvarts and Bakker's 2019 history of the metaphor found the word in Vygotsky's notebooks, but he did not develop scaffolding as a teaching concept. So scaffolding is one way to help someone inside their zone, and the zone is the gap that the help reveals.

The zone is not a score band or a fixed level

Some tools turn the ZPD into a range of test scores or reading levels. Vygotsky did put numbers on it, saying the zone could be expressed by the number "4" for one child and "1" for the other (Thinking and Speech, p. 208). Those numbers illustrate a difference between two levels for one person, and they change between his texts. The zone belongs to a person and a kind of task at a point in time, and it moves: "what is in the zone of proximal development today will be the actual developmental level tomorrow" (Mind in Society, p. 87). Chaiklin adds that "Even the classic definition refers to levels of development, not tasks", and that there is no reason to think the size of a zone is fixed.

Not three core components, and not five stages

The ZPD is often presented as three core components: a more knowledgeable other, scaffolding and social interaction. Vygotsky's own building blocks are different: two developmental levels, the maturing functions between them, the limits of imitation and the use of the zone for diagnosis. Nor does the zone come in five stages. The texts that define it describe one distance between two levels.

Dynamic assessment: testing with help

Vygotsky's argument was about assessment before it was about teaching, and dynamic assessment is the line of work that took it literally. In their 2005 paper on dynamic assessment, Matthew Poehner and James Lantolf note that it has "its roots in Vygotsky's concept of the ZPD", but that "Vygotsky himself did not use the term DA" (p. 233). Reuven Feuerstein's Learning Potential Assessment Device, set out in a 1979 book, built a clinical version around an adult mediator who works through tasks with the learner.

Two broad styles have grown up since (p. 239):

• Interventionist dynamic assessment treats the zone as a difference score. It uses a pretest, some teaching and a post-test, or hints chosen from "a prefabricated menu" during a test.
• Interactionist dynamic assessment works more like a conversation, with the help shaped to each learner's responses as they go.

For workplace training, the most portable tool from this work is the graded hint: help that starts implicit and becomes explicit only when needed, with a record of how far each person had to go. A clear example comes from a writing tutor in Delaware.

Case study: Graduated feedback for language learners in Delaware

This was a small tutoring study designed around the ZPD, in a university English language institute rather than a workplace. Three ESL learners in an eight-week reading and writing course at the University of Delaware, whose first languages were Japanese, Spanish and Portuguese, met a tutor each week to work on errors in their own essays.

• What they did: In one-to-one tutorials of 30 to 45 minutes, the tutor started with the least help possible and moved to more explicit help only when needed. Ali Aljaafreh and James Lantolf's 1994 paper set this out as a 13-level scale, from asking the learner to find and fix errors alone (level 0), through "Is there anything wrong in this sentence?" (level 3) and naming the kind of error (level 6), to giving the correct form (level 10) and examples (level 12). Help was "offered only when it is needed, and withdrawn as soon as the novice shows signs of self-control" (p. 468).
• Evidence and result: Over the weeks, the learners needed less explicit help with the target forms: articles, tense, prepositions and modal verbs. One learner who first needed explicit help with modal verbs later used them correctly alone in her essays. The authors argued that a learner who can correct an error after an implicit hint is "developmentally more advanced than one who needs direct and explicit feedback for the same property" (p. 480).
• Limitation: Three learners, a tutor who was also the researcher and no comparison group. The scale was drawn from the tutorials rather than set in advance, and the errors were surface grammar. It shows what progress looks like, not that the method beats others.
• Lesson: How much help someone needs is itself a measure of progress. Write hints from implicit to explicit, give the least that works and note which step each person needed. Someone who needs less help each week is improving before they are error-free.

The worked example: new receptionists at a vet clinic

Here is the hypothetical example I'll use from here on. A small-animal clinic with two vets has one front desk, and six new receptionists in their first month answer calls from pet owners. They start from very different places. Two came from a pet supplies store, so they know animals and owners but not clinical urgency. Two came from a GP clinic's front desk, so they know phone triage for people but not for animals. Two are school leavers with no customer or animal-care work. All six have done the clinic induction and the booking system training, and none of their previous jobs shows they can judge urgency for an animal.

The task is first-contact triage: decide the urgency from the clinic's triage list, choose the next step and tell the owner what happens next, without giving medical advice. The simplified list below is written for this training example. A real clinic must use its veterinarian-approved triage protocol, including escalation for unclear symptoms. The example has three levels:

• Emergency, come in now and alert the duty vet: difficulty breathing, collapse, a male cat straining in the litter tray with little or no urine passing, a dog with a swollen belly retching without bringing anything up, suspected poisoning (such as chocolate, rat bait, or lilies for a cat) and bleeding that will not stop.
• Same-day appointment: vomiting or diarrhoea in an animal that is otherwise bright and drinking, limping while still putting weight on the leg, or not eating for one day.
• Routine appointment: vaccinations, nail trims and problems that have lasted more than a week in an animal that is otherwise well.

Receptionists never give medical advice and never tell an owner an animal can wait if any emergency sign is present. If unsure, they put the call through to the duty vet, who is told about every emergency and is the only person who can say an animal can safely wait. Questions about fees go to the practice manager. Each new receptionist is paired with a senior receptionist for the month.

The course uses four calls with fixed facts:

• Call 1, the first attempt with no help: a 10 kg terrier ate most of a 100 g block of dark chocolate about 30 minutes ago and seems fine so far. It is an emergency: come in now, alert the duty vet and bring the wrapper.
• Call A, practice: a 4-year-old male cat keeps going to the litter tray, straining, with only drops of urine, and crying. The owner thinks he is constipated and asks for an appointment tomorrow. It is an emergency, whatever the owner thinks the cause is.
• Call B, practice: a 6-year-old Labrador vomited twice this morning and is now bright, eating grass and drinking normally. The owner asks whether to come in and whether to give a tablet from the cupboard. It needs a same-day appointment, no advice about the tablet, and a call back straight away if the dog keeps vomiting, collapses or its belly swells with retching.
• Call 2, the final call with no help: a 3-year-old Great Dane ate dinner an hour ago, is restless and drooling, keeps trying to vomit with nothing coming up, and its belly looks bigger. It is an emergency.

By the end of the first month, each receptionist should triage a typical call with the senior receptionist checking the note, alert the duty vet on every emergency sign, give no medical advice and know which help they still reach for. The outcome does not include clinical judgement. A course also cannot show how anyone handles live calls. The example suits the ZPD because the six may all miss Call 1 in a simulated exercise and still need very different amounts of help on Call A, which echoes Vygotsky's assessment example without measuring development.

How to design training around the zone of proximal development

Vygotsky described the zone, not a method for teaching in it. Chaiklin notes that he "does not seem to have any systematic principles, methods or techniques" for how the helping should be done. So this workflow is my own design. It draws on Vygotsky's two levels, dynamic assessment and the graded hints from the Delaware study, and each step uses the front-desk course.

1. Start with an attempt alone. Use a simulated task before teaching and record what each person does. Never withhold help on a live clinical call to test learning. Call 1 is the unassisted attempt: an open-text triage of the chocolate call, then a model note to compare against, with no score. This records independent performance on this call, not a developmental level.
2. Write the help as a ladder, from implicit to explicit. For this example, use four steps: a nudge, a pointer, a worked part and a full demonstration. Adapt the number and kind of hints to the task. The demonstration works the way social cognitive theory describes learning from a model, and it comes last.
3. Offer the least help first, and keep the rest out of sight. Learners try, then open one step at a time. If every step is visible at once, everyone reads the answer and you lose the record of who needed what.
4. Record the step each person needed. A receptionist who needed only the nudge on Call A is closer to working alone than one who needed the full note, even if both end with the right answer. That record describes assisted performance on the call. It does not measure a level of potential development.
5. Pair each learner with a more capable colleague. The senior receptionist plays the "more capable peer" that Vygotsky described. Give them the help record, so their time with the new starter goes on the calls that needed the most help.
6. Take help away as people improve. Offer the ladder again on Call B and expect people to need lower steps. Wood, Bruner and Ross's tutor already worked this way: "The child's success or failure at any point in time thus determined the tutor's next level of instruction" (p. 92).
7. Finish with a new task, done alone. Call 2 is a different emergency with no help offered. It checks independent performance on a new emergency call, without establishing a general developmental change, and it is a summative check, while the help record works as formative feedback along the way. Our guide to formative vs summative assessment covers the difference.
8. Read the record, not a total. The step someone opened is a rough signal of where they are, not a measure of development. Use it to decide what to practise next and when the senior receptionist can hand over more calls.

The help ladder for Call A

Call A has a built-in trap: the owner thinks the cat is constipated and asks for tomorrow. Here is the ladder for it, from the lightest help to the heaviest.

Withdrawing help in stages has been tried in a small hospital study in China, with a caveat about what it can prove.

Case study: Anaesthesiology residents at a hospital in Ningbo, China

This was a small controlled comparison of a combined package, so the ZPD cannot take the credit alone. At Ningbo Medical Center Lihuili Hospital, anaesthesiology residents struggled with complex, high-risk cases, and ordinary case discussion gave them little structured guidance.

• What they did: The intervention group had four weekly sessions of 1.5 hours on anaesthesia for one complex heart condition, which the authors describe as support "within learners' Zone of Proximal Development (ZPD)" with "graduated withdrawal of support". Week 1 was a guided analysis of the case, week 2 a group management plan refined by the instructor, week 3 simulated crises with role play and week 4 presentations with structured feedback. The comparison group had lectures on the same case with questions and answers.
• Evidence and result: In Zhou and colleagues' 2025 paper in BMC Medical Education, clinical reasoning observed in simulated crises scored 83.58 against 74.17 at week 4 and 89.08 against 68.17 at week 8, with 6 residents in each group. The knowledge test showed no clear difference at week 4 (99.17 against 96.67), with both groups close to full marks.
• Limitation: The paper calls it a randomised trial, but the groups were allocated by campus and had different teaching teams, so the site and the teachers are tangled up with the method. Six per group is very small, the scores come from simulation rather than patient care, and the authors say the design compares lectures with the whole case-based package "rather than isolating the specific impact of scaffolding".
• Lesson: For a high-stakes task, give heavy guidance early and take it away in stages as people move to simulated practice and then defend their own decisions. Compare against your current format, and measure what people do in the task rather than how satisfied they were.

The more capable peer does not have to be senior, either. In a 2004 programme in North West England, owner-managers of small businesses met in action learning sets of about six to question each other on real business problems. Anderson and Gold's 2009 evaluation, built on observation, 19 learning journals and 26 interviews, describes one owner who took "a much more focused and targeted approach to finding new customers" and joined a speaking club he said he would once have "avoided at all costs". It rests on two hand-picked stories and self-report, but it shows that peers who ask probing questions can stretch experienced people, not only new starters.

What course requests leave out

To see whether people brief courses with the zone in mind, I looked at course requests typed into Coursebox in the 90 days to 7 October 2026: one per person, from English-language browsers, with no uploaded file. Of 266 prompts, 104 were course requests. Two independent AI readers each classified all 104 against written definitions, and every positive call was re-read by hand. Counts combine the two readers' classifications and remain estimates. Checking positive calls does not establish how many requests were missed.

Six of the nine requests for rising difficulty used a label such as "beginner to advanced", which describes the range of the content, not help that responds to the learner. Two requests classified as workplace courses described a mixed group of learners, such as "current and aspiring" practitioners, and neither asked for anything other than one route for everyone. That resembles the assessment problem in Vygotsky's example: people who look alike on paper can need very different amounts of help. The sample is small, it covers requests rather than finished courses, and trainers may well check people's level or coach them away from the course. Still, if you want a course that works with each learner's zone, write the first attempt, the help steps and the record into the brief yourself.

ZPD vs scaffolding, i+1, flow and cognitive load: which idea to use

These five ideas are often used as if they were one. They answer different questions. The table is my synthesis, not a ranking, and I found no study comparing them head to head.

Scaffolding is the closest neighbour. Wood, Bruner and Ross warned that "The major risk is in creating too much dependency on the tutor" (p. 98). A 2010 review by Janneke van de Pol and colleagues named "contingency, fading, and transfer of responsibility" as the three key features of scaffolding (p. 271), and states the rule that the tutor "increases support to the extent that the learner fails and decreases support to the extent that the child succeeds" (p. 281). For the front desk, the ZPD tells you to record what each receptionist needed, and scaffolding tells you how to write the four steps and when to take them away.

Stephen Krashen's i+1 sounds similar. In his 1982 book he argued that "We acquire, in other words, only when we understand language that contains structure that is 'a little beyond' where we are now" (p. 21). That is about understanding input in a second language, and Dunn and Lantolf argued in 1998 that the two concepts are "unrelatable". Flow is different again. Mihaly Csikszentmihalyi wrote that "Enjoyment appears at the boundary between boredom and anxiety" (Flow, 1990, p. 52). Flow describes how an activity feels while you do it; the ZPD describes what someone can do with help that they cannot yet do alone.

Cognitive load theory explains why novices need worked examples and why guidance should fade, which helps when sizing each step of the ladder. Cognitive apprenticeship gives the senior receptionist a method for modelling and coaching, and situated learning explains why new starters learn most by taking part in the real work at the desk.

Vygotsky is also often set against Jean Piaget. Piaget described children's thinking developing through stages as they act on the world, while Vygotsky argued that learning with other people runs ahead of development and pulls it along: "an essential feature of learning is that it creates the zone of proximal development" (Mind in Society, p. 90). Our guide to constructivism covers both, and our comparison of behaviourism, cognitivism and constructivism places them in the wider history of learning theory.

Criticisms and limits of the zone of proximal development

Five criticisms come up in the research, most of them about how the idea gets used.

• It is vague and hard to measure. Chaiklin quotes James Wertsch's 1984 warning that the concept risks being used "loosely and indiscriminately", and notes that Vygotsky gave no systematic method for assessing a zone. A 2021 paper by Poehner and Lantolf reports the argument, from Valsiner and van der Veer, that "direct, empirical study of the ZPD is not possible as it can only be inferred retrospectively" (p. 31).
• A small idea became a big one. Margaret Gredler argued in 2012 that "inaccurate information about a minor discussion, the zone of proximal development (ZPD), attracted attention early on and became identified as a major aspect of his theory".
• It is stretched well beyond child development. Chaiklin calls the belief that the zone applies to learning of every kind the "generality assumption". For teaching a specific skill he suggests other terms, such as "assisted instruction, scaffolding". Using the zone for adult job skills, as I do here, is a choice to make knowingly.
• It gets collapsed into scaffolding. Xi and Lantolf argued in 2021 that any similarity between the two "is at best partial and at worst superficial", and Addison Stone noted in 1998 that the scaffolding metaphor can make help "too other-driven or one-sided".
• Testing with help predicts better only under some conditions. Caffrey, Fuchs and Fuchs's 2008 review of 24 studies found dynamic assessment predicted achievement better under some conditions, such as with students with disabilities rather than typically achieving students, and the samples were school children. Elena Grigorenko's 2009 review concluded that clinical approaches such as Feuerstein's "have not been proven effective empirically in the research literature". Testing with help also takes longer per learner than a standard test.

The research on help itself is small, too.

Van de Pol and colleagues found that the results "suggest that scaffolding is effective" (p. 271), but called for more research and named measurement as the main challenge. The workplace evidence is thinner again. Every controlled study I found bundles ZPD-based help with other changes, such as case discussion, simulation or small groups, so none shows what the ZPD adds on its own, and I found no verified case in trades, sales or customer service that names the ZPD and reports outcomes. Treat a help ladder as a sensible design to pilot and check, not a proven method.

Using the zone of proximal development with an AI course builder

On 8 October 2026 I briefed Coursebox to build the front-desk course. Coursebox is our product, so weigh my account with that in mind. The brief described the six receptionists, the triage list and the four calls, and asked for eight sections in order: a short "Before you start", Call 1 as an open-text attempt with no help, a walk through the triage list, practice calls A and B with four help steps each hidden until opened, Call 2 with no help, a guide for the conversation with the senior receptionist, and a wrap-up last. I asked for open-text answers instead of quizzes, and about 45 minutes.

The first outline missed the part that makes this a ZPD design. It came back with 5 sections instead of 8. Both practice calls were squeezed into one section, Call B and the senior receptionist conversation became quizzes, and the four help steps did not appear anywhere. As outlined, every receptionist would follow the same route at the same pitch, with no way to see who needed how much help.

One follow-up message fixed it, after the builder paused once to ask how to size the pages. I asked for the eight sections back in order, with each practice call split into an open-text triage, then a page with the four help steps as separate click-to-reveal items, then an open-text revision recording which step the learner needed, and no quizzes anywhere. I reloaded the outline and checked it: 8 sections, 15 pages and no quizzes. Then I opened the generated pages for Call A. On "Open four help steps: Call A", the nudge, the pointer, the emergency line and the call note each sit in their own collapsed panel, and I checked that each stays hidden until the learner opens it. The learner writes a triage on the page before and revises it on the page after.

Two steps came back thinner than I briefed: the worked part gives the matching emergency line without marking the key words, and the demonstration is a short call note rather than the senior receptionist's full note. The course also runs to about 1 hour 34 minutes against the 45 I asked for, so it needs trimming. I checked the outline and these pages, not a run with receptionists, and the record of which step someone opened is only useful if the senior receptionist reads it. This course test did not assess how a receptionist handles a worried owner on a live call. With the ladder in place, though, the course now asks for an attempt alone before any help, offers help that steps up only when the learner opens it, and gives Call 2 its own "Your Call, No Help" section.

The other half of the work happens at the desk, where the senior receptionist listens to real calls and decides when to hand over more.

Putting the zone of proximal development to work

The zone of proximal development answers a question that no course brief in our sample asked: what can this person do with help that they cannot yet do alone? Vygotsky's answer was to measure both, and to aim teaching at the distance between them. At work that means an attempt alone, help that steps up only when needed, a record of how much each person used and a new task done without help at the end. Treat the zone as a lens for diagnosis, not a score. It belongs to one person and one kind of task, and it moves as they learn.

Start with one task you train. Write four hints for it, from a nudge to a full demonstration, and the next time a new starter gets stuck, give them one step at a time and note the step where they got there.