Article
Education & Watchmaking: Quality, Lost in Translation
I'm an educator turned edtech founder. For the better part of a decade, my work has orbited something the field calls "education quality" - a phrase I actually try not to use, because it's too broad to pin down. I've led assessment-system development, curriculum integration and development, standards localization, accreditation project management, educator CPD, and sat in plenty of rooms where people argue about what "good" actually means. And all through it, I've been married to a quality assurance manager from the Swiss watchmaking industry. This turns out to be a recipe for some very strange dinner conversations.
My husband is a watchmaker turned quality assurance manager. For twenty years his job has been deciding whether a movement is good enough to carry the brand. Everything in his world is precision and detail, and the prestige names he works for come with expectations to match, customers who demand the same perfection from the movement they'll never see as from the watch they will. He thinks in microns, with hands like a surgeon's and a mind that knows the hundreds of tiny parts of a movement inside out. Two decades of this gets into a person: he's so deeply shaped by the craft that he's developed a quiet superpower, he can tell you the time without glancing at any watch, eight times out of ten, to the exact minute, and rarely off by more than three.
We talk about education quality constantly, and agree on maybe half of it. The line of his that stuck with me is one he didn't coin but lives by: quality is simply expectations met, what the quality literature since Joseph Juran has called "fitness for use," the customer's expectations satisfied. A watch is good when it delivers exactly what its owner was promised, and at his end of the market that's far more than keeping time; it's the beauty, the heritage, the weight on the wrist, the pride of owning something built to outlive you. By this measure a modest watch that keeps its modest promise is high quality, and an expensive one that disappoints its owner is low, price never enters into it, only whether the promise was kept. The promise differs from watch to watch, but for any given watch it's knowable and agreed, fixed before a single part is cut. Match it and you have quality, miss it and you don't. Clean and workable. So how about education? Or, let's not challenge ourselves that far, how about a single school? This is where that clean definition quietly comes apart, because the moment you ask whose expectations, and of what, the whole thing splinters.
My vantage was never the front of a single classroom. I spent some time with students, but far more with the people a step back from them - school leaders, the heads of education groups, teachers - working across institutions rather than inside any one room. I've watched "quality" get argued over in leadership meetings, accreditation reviews, and staff rooms. But all of it has to land somewhere concrete: real classrooms, real students, no two alike and none of them guaranteed to be paying attention that day.
For years we'd compare notes over dinner, and slowly I realized something: my husband and I use exactly the same words - quality, standards, improvement, consistency - and we mean almost nothing in common by them. The vocabulary of quality was forged in places like that watch factory. Education imported plenty of it, Total Quality Management swept through schools in the late 1980s, and the vocabulary still lives in accreditation, quality assurance, and school-improvement plans. I leaned on it as much as anyone. The borrowing hides a truth that every teacher feels in their bones, and that I kept bumping into from the quality side of the desk: managing quality in a school is nothing like managing it on a workbench. It's harder. Not because teachers are worse than watchmakers, but because of what we're trying to make, and how we have to make it.
The best way I've found to explain this to my husband uses a tool from their world. So let me steal it.
The fish on the napkin
The tool that finally helped me explain my problem comes from his side of the table. My husband uses the fishbone diagram to trace why a watch movement runs badly, the cause-and-effect diagram invented by the Japanese engineer Kaoru Ishikawa, a staple of quality work everywhere. You put the problem at the head of a fish skeleton, and six bones feed into the spine, each a family of causes, and, handily, they all start with M: Man (people), Machine (tools), Material (your inputs), Method (the process), Measurement (how you know things), and Mother Nature (the surrounding environment). One evening, comparing notes, we sketched one on a napkin between us.
"Whenever a movement runs badly," my husband said, "the cause is always on one of these bones."
"Right," I said. "Let me show you why that same fish drives me quietly insane."
So we went bone by bone. Here's what we found, and in every case, the thing that makes the bone easy to manage in a watch factory is exactly the thing I don't have.
Man: my raw material argues with me
Ask quality professionals which of the six M's gives them the most trouble and many name the same one: Man. People are the hardest variable to control even in a watch workshop, and in a school they're very nearly the whole system.
In the workshop, "Man" is the people doing the work: skilled, trained, standing on one side of the bench while the watch sits on the other. Nobody confuses the two. A school smears that line into mush, because three different sets of people shape every result.
The teachers. Teacher quality is the most-studied lever in education, and it matters enormously, but teaching isn't standard work. The same lesson plan lands differently in every room and every hour. The craft is the art of not being interchangeable: reading the room, finding the one analogy that finally makes it click. A factory wants assemblers it can swap like batteries; the best teachers are valuable precisely because you can't.
The students. This is the one that ends the argument: the raw material is a person, and the person gets a vote. A watch component never shows up exhausted, anxious, or convinced the whole thing is pointless, and it doesn't have to agree to become a watch. A student does, and has to. Learning isn't done to a child; it happens only with them, their attention, their effort, their willingness. You cannot promise an output when half the work is done by someone you don't control and have to win over every morning.
Everyone else. Parents, leaders, support staff, and above all the other students, classmates are part of the machinery acting on each child, and it's a different set in every room.
So the "Man" bone isn't one thing; it's a crowd. And one member of that crowd is somehow the raw material, the worker, and the finished product all at once.
Machine: tools that quietly change the job
In watchmaking, machines are the most controllable bone of the lot - lathes, cleaning baths, timing machines, all calibrated and held to tolerances finer than a human hair. When one drifts, you know, and you fix it. My "machines": the tech, the building, the kit, misbehave in far more interesting ways.
No fixed job. A lathe removes a precise sliver of metal, the same job tomorrow. A learning platform, an interactive whiteboard, an AI tutor has no single job; it reshapes the whole lesson around itself. When my school gave every student a tablet, the change wasn't "the same teaching, now digital." Reading moved from books to screens, homework became apps, the teacher turned into part-time IT support, and attention splintered against every notification. We didn't sharpen a process, we changed what teaching and learning were. The machine rewrote the definition of the product. A lathe never does that.
A murky payoff. The link between classroom tech and actual learning is stubbornly unclear. More technology doesn't reliably mean more learning, sometimes it helps, sometimes it distracts, sometimes it just moves money from teachers to vendors while results sit still. Generative AI cranks the tension to eleven: the same tool is a tireless tutor or an essay-mill nobody thinks through — and because it now generates work and makes judgments itself, "does it run" is no longer the same question as "is what it produces any good." My husband can measure exactly what a tool contributes to a watch; the gauge tells him. I can't isolate what any single tool adds to learning, because its value depends entirely on how it's used.
Not one machine, a stack. Education's "machine" is now dozens of tools at once, a learning platform, a student-information system, assessment and analytics, adaptive courseware, an AI layer on top, around fifty a year for the average teacher. The pressure is sharpest in the private and international schools I work with, where fee-paying parents expect the newest thing and no budget brake slows the buying. My husband's toolset is deliberately controlled, every tool earning its place; an edtech stack just accretes, and since each is marketed as the all-in-one that does everything, the layers overlap rather than fit together, leaving schools unsure which tool is even for what. The most advanced stack on the market can still teach worse, if it pulls attention from the few things that actually move learning.
Unequal by default. Watch factories are unequal too, a luxury manufacture and a budget workshop own wildly different equipment. But each is judged against the watch it actually set out to make; nobody holds a fifty-dollar quartz line to a chronometer's standard. Schools get the opposite treatment. Their "machines": labs, fast Wi-Fi, a warm dry building, differ just as much. But a child in the under-resourced school is no less deserving than one in the lavish one, so we rightly refuse to lower the bar for them, and then blame the school itself for a gap it never created, instead of closing it.
Material: I don't get to reject the inputs
This is the bone where the whole comparison snaps, and it's worth saying plainly, because the polite version has done real harm.
My husband inspects every incoming batch - brass, steel, jewels, oils - against a specification, and rejects anything out of tolerance on the spot. You never build a fine watch on a flawed base. The entire logic of quality control starts from controlling your inputs.
You can't reject yours. Put a child where the brass goes and the logic curdles, which is exactly the point. My "material" is children: human beings, every one different, arriving with no spec sheet and no single standard they can fairly be judged against. I take who walks through the door, ready and not-ready, fluent and still learning the language, fed and hungry, advantaged and traumatised, and then I'm judged on what I did with all of them.
So output mostly measures the intake. Pass rates and average scores mostly reveal who walked in, not what the school added. Dazzling results may just mean dazzling kids; grim numbers may hide heroic work with children who arrived years behind. Judge a school on output alone and you reward picking good intake and punish the hardest work in the system.
And it tempts you to game the intake. Selective admissions, quietly nudging out the difficult kids, counseling "weak" students off the register before the exam that counts, pouring energy into the kids right at a grade boundary, it's the same move my husband makes at the loading dock, fixing the output by controlling the material. It reliably lifts the numbers and just as reliably betrays the point of a school. Worse, the selection rests on the crudest read of a child: a test score, a single interview, while the things that matter most (talent in an untested field, character, potential, the open-ended spark) are the hardest to measure and the first to be missed.
The only fair measure is distance traveled. Quality here isn't the state of the finished thing but the distance from where a student started, what's called value added. My husband never thinks this way, a finished watch is judged against an absolute standard, not the lump of brass it used to be. For me it's the only fair measure, and brutal to pin down: you have to know the starting point, separate the school's contribution from everything else in a child's life, and wait years to learn if it worked.
And the child has to be seen whole. Distance traveled in what, exactly? Not just an exam score. A child grows along many axes at once, knowledge, but also character, creativity, confidence, social and emotional maturity, the appetite to keep learning — and an honest measure has to see the whole person, not collapse them into a single grade. A watch can be summed up in one number; a child never can. Holistic assessment is slower, messier, and impossible to rank cleanly on a spreadsheet, which is exactly why systems shy away from it, and exactly why it matters. Which lands us on the next bone.
Method: a recipe nobody agrees on
Method is the beating heart of quality management. The founding insight, from Shewhart through W. Edwards Deming, is that quality lives in the process, not the final inspection. Find the best-known method, standardize it, squeeze out the variation, measure, improve in disciplined loops. It's how a watch movement goes from "good" to "exceptional."
My relationship with the word "method" is complicated, in three ways.
No one agrees on the best one. Phonics or rich stories? Procedure or concepts? Lecture or flip the classroom? And beyond the old debates, schools pile up whole branded philosophies: inquiry-, project-, concept-, evidence-based learning, each with advocates, none settled. Some have real evidence behind them, but education is also a contested social project: people don't even agree on its purpose: good citizens, employable adults, curious humans? My husband knows exactly what a good watch is. Society doesn't agree on what a good education is, and that disagreement is legitimate, not a defect to grind out.
The best method changes with every child. Education is full of interaction effects: what's magic for one kid flops for the next. Watchmaking wants to crush variation around a single standard; good teaching often means generating it, differentiating, adapting, meeting kids where they are. A quality system tuned to minimize variation can stamp out the exact responsiveness learning needs. The thing my husband would flag as a defect is sometimes just good teaching.
The feedback loop runs in years, so the process warps. My husband tests a movement and knows within fifteen days whether it passed; my "product" takes years to form, and its most important features - did this person become curious, capable, decent? - can't be seen until long after they've left. With no real-time signal, schools lean on proxies and then standardize around them, and that's how you get teaching to the test, a shrinking curriculum, and the quiet abandonment of everything valuable but unexamined. A timing machine doesn't reinterpret its targets to hit the number. A school does, all the time.
Measurement: when the ruler argues back
Ishikawa gave measurement its own bone for a sharp reason: a bad gauge can invent a problem that isn't there or hide one that is. So my husband calibrates obsessively, and trusts the numbers before acting on them. Settled technical discipline.
And here's the reality I keep running into: in education, one gauge has become so entrenched that almost everyone treats it as the truth. "Student success" has quietly collapsed into grades, "school success" into the aggregate of them, and parents reach for that number first, because it's the one the whole system trained everyone to read. So the factory way of managing success never has to be argued for, it's the default. "But schools track so much more now," people tell me, and they do; the question is at what scale it counts, and the exam is still the gatekeeper. Just missing a key grade threshold measurably lowers a student's odds of a degree and dents their later earnings: one number, or a hundred, far too small a thing to weigh a life on. All of which would be fine if the gauge were trustworthy. It isn't, in three layers.
Are we even measuring the right thing? Education isn't a process with one clean output; it's the work of managing deep uncertainty, a developing human shaped by a vast, shifting tangle of variables. Its real aims, understanding, judgment, creativity, character, the appetite to keep learning, are abstract, slow, and resist direct measurement, so we fall back on stand-ins: test scores, grades, completion, satisfaction surveys. Each catches part of the target and misses the rest, and the missed part is often what mattered most: a reading test captures decoding but says nothing about whether a child fell in love with books. Manage by the proxy and you quietly redefine quality as the proxy; the map eats the territory. A fine watch has no such gap, held to within two seconds a day, that one number is the thing my husband cares about, exactly and completely.
The ruler talks back. This is Goodhart's Law: the moment a measure becomes a target, it stops being a good measure. A timing machine doesn't change its behavior because you're watching it. A school absolutely does. Attach high stakes to a metric - funding, reputation, jobs, league-table rank - and the whole system reorganizes to game it, often in ways that hollow out the goal. Test scores climb while learning doesn't. Graduation rates rise while standards fall. Satisfaction scores improve as courses get easier. In education the measuring instrument isn't a quiet observer, it's an intervention that changes the thing it's measuring. This is probably the single biggest reason factory-style metric management keeps backfiring in schools.
The signal is buried in noise. Even an honest, valid measure is drowning in static. One year's results for a school reflect that specific cohort, the local economy, a flu season, a charismatic teacher who's since left, a tweaked exam, and plain luck, far more than any stable thing called "quality." A small school's ranking can lurch wildly year to year on pure chance. React to that noise as if it were signal and you commit exactly the over-control Deming warned about: meddling with random variation makes the system worse. We do it constantly, because the political demand for accountability won't wait for the statistics to settle.
Stack the three and you reach something uncomfortable: my measurement system isn't a neutral instrument I can calibrate and trust. It's contested, reflexive, noisy, and partial, and the harder I lean on it, the more it distorts the very thing I want to improve. The answer isn't to cling to the old number because change is hard; there are better ways to see a child, and we've mostly lacked the nerve to try them. As Ken Robinson argued, education needs not evolution but revolution.
Mother Nature: you can't build a clean room around a childhood
The last bone is the environment. In watchmaking it's dust, humidity, temperature, vibration, the ambient stuff that ruins precision, and the answer is famous: wall it out. Watches are built in sealed clean rooms so the environment can't reach the work. A school can't wall the world out, and shouldn't want to. The environment isn't noise to exclude from learning; it's half the content and most of the cause.
A child's life. A huge share of the variation in outcomes traces to things outside school entirely: family income, parents' education, neighborhood, health, nutrition, whether home is stable or frightening. A child who shows up hungry, sleepless, or scared can't learn well however good the teaching, and no school can climate-control the conditions of a life. My husband's materials never go home each afternoon and come back changed.
Policy and politics. Schools run inside a permanent churn of shifting regulations, funding cycles, and political fashions, a reform launched under one government reversed by the next before anyone can tell if it worked. A watch factory gets long stretches of stable conditions to perfect a movement; a school rarely gets a decade of peace to let one idea mature.
The climate of the room. Safety, belonging, relationships, whether a kid feels known, these aren't background to learning, they're among its strongest drivers. And unlike temperature, no thermostat sets them; they're produced continuously by the same hard-to-direct humans from the very first bone.
And notice what "good" even means. A clean room is judged by how little happens in it: no dust, no movement, no disturbance. A learning environment is the opposite: not the most pristine room in the building but the one that best supports learning, which is noisy, social, and alive. The very things a watch factory exists to eliminate - activity, unpredictability, life - are the things a school has to invite in. So the relationship inverts: a factory isolates its environment; a school has to engage with it. And the asymmetry is brutal, a school gets a child for a few hours of lesson time while everything else, home, health, friends, the whole weather of a life, works on them the rest of the day. Trying to wall that out isn't just impractical. It's a category error.
What the fish on the napkin actually said
By the end of the bottle, my husband and I were looking at the same little drawing and seeing it completely differently. In every bone, what makes quality manageable on his workbench is what a school lacks: its people co-produce the result; its machines redefine the product rather than refine it; its material can't be turned away; its method is contested; its measurement corrupts the moment it becomes a target; its environment can't be walled out. Point for point, education is the factory's inverse.
That's why importing the watch-factory model wholesale produces such a recognizable mess: proxies worshipped over purposes, gamed metrics, a narrowing curriculum, quiet selection of easy intakes, overreaction to noise, and the slow demoralizing of the people the system most needs. Schools aren't badly run. It's what happens when you run a control system built for a closed, repeatable, input-controlled process on one that's open, human, co-produced, and value-laden.
But, and this is what my husband insisted on, and he's right, none of it means quality can't be managed in education. The quality tradition has a great deal to offer. It just has to be translated, not transplanted. Ken Robinson drew the line exactly: education has to move from a manufacturing model to something closer to agriculture, because "human flourishing is not a mechanical process; it's an organic process… all you can do, like a farmer, is create the conditions under which they will begin to flourish." That's the task - building the conditions, not machining the part. Five principles fell out of our napkin:
Record the journey, don't rush to a verdict. A school's real effects can surface a decade later, long after any report card, and much of what matters most was never measurable in the first place, only noticeable and recordable. "Outcome" is too brutal and too final a word for something this slow and this human. So the work isn't to crunch a child, or a school, into a single number; it's to keep a faithful record of the journey, the work produced, the growth witnessed, the conditions created, the evidence gathered year on year. Quality lives upstream anyway, in what a school builds: relationships, the breadth of what it offers, whether children feel safe, known, and want to be there. Hold the record, resist drawing the conclusion early, and the picture stays fairer and more honest than any verdict.
Treat measurement as fallible evidence, and keep the stakes proportionate. Goodhart's Law - the moment a measure becomes a target, it stops being a good measure - isn't a problem you solve; it's a condition you manage. The higher the stakes on one number, the more it lies to you. Several balanced indicators held loosely distort less than one sharp number held in a death grip. And the richest evidence is rarely a number at all, it's the record of the journey: a portfolio, a body of work, the trail of how a child actually grew, which any single score will always flatten. The same goes for teachers: their craft no more fits in their students' test scores than a child's growth fits in a grade, judge them, too, on the fuller record of their practice. Good accreditation already works this way, weighing a school holistically, across many standards and a whole body of evidence, never on one number. Parents deserve to be brought into that same view: helped to judge a school by the whole picture, not the league-table line they were trained to read.
Cut the bad variation, protect the good. A factory tries to make every watch identical; a school can't, and shouldn't. Some variation is genuine defect: chaotic, unsafe, or incompetent teaching, a toxic culture, broken communication, no clarity of purpose, and should go. But some is the entire point: a teacher adapting the same lesson for a child who's lost and a child who's bored. Standardize everything and you kill the second kind along with the first. The goal isn't sameness, it's removing the harmful inconsistency while protecting the responsive kind. That takes mindfulness, not a blanket rule: someone close enough to tell the difference, case by case, between a teacher improvising brilliantly and one simply winging it.
Invest in people, because quality walks in with the teacher. You can't install quality in a school; it arrives with the people. So the moves that matter most are human ones: real training, time for teachers to plan together and watch each other teach, mentoring for newer staff, and the freedom to adapt a lesson instead of reading from a script. It also takes trust - a culture where a teacher can admit "this class isn't working" without being blamed for it. (W. Edwards Deming, the quality pioneer, called this "driving out fear": frightened teachers hide problems instead of fixing them.) And as Rita Pierson said, "every child deserves a champion, an adult who will never give up on them, who understands the power of connection, and insists that they become the best that they can possibly be." Why? Because "kids don't learn from people they don't like." You can't script a champion into being, you can only hire good people, help them grow, and trust them.
Engage the environment instead of pretending to seal it out. Because so much of a child's progress is set beyond the school gate, a quality strategy that ignores health, nutrition, family, and stable policy is measuring the weather and calling it seamanship. So the best schools work with the wider community rather than around it: strong home–school relationships, pastoral and wellbeing support, breakfast clubs, safeguarding, the often invisible work of spotting and protecting the children most at risk of harm or neglect, and real partnerships with the families and services around each child. A school can't control a child's world, but it can reach into it, and the ones that do tend to move children furthest.
Ishikawa built his fish to help engineers stop guessing and start asking, systematically, where a problem really comes from. I've used it here to explain why education will never be that clean, and why that isn't a failure but the whole point. Education isn't a defective watch; it's a different kind of thing entirely, open where the workbench is closed, human where it's mechanical, contested where it's settled, slow where it's fast. So we washed up the glasses, and my husband went back to perfecting hairsprings while I went back to the question this whole conversation kept circling: if education can't be machined like a watch, how do we honour what it actually is? That question is why I'm building Novana, not another gauge that crushes a school into a single score, but a way to hold the #evidence of the journey itself: the record of how a school grows over time, the conditions, the data, the story, kept faithfully rather than flattened into a verdict. Watchmaking taught my husband to measure what's certain; education is teaching me to record what isn't, and to trust that the fuller, slower, more human picture is the truer one.
And we'd finally found the one thing we completely agree on: quality starts with #clarity, and the courage to be honest with ourselves. Before you can manage it, improve it, or measure anything, you have to face what kind of thing you're actually making. Read carefully, the little fish on the napkin says exactly that.
If any of this resonates, or if you think I've got it wrong, I'd genuinely like to hear from you. We're building Novana in the open, and the sharpest thinking keeps coming from conversations with educators, school leaders, and parents who care about getting this right. Connect with me, and let's talk.