The PISA results came out this month, and like most people in this industry I have spent more time with them than I meant to.
Let me say at the outset what this is not. It is not an analysis. There are people far better qualified than I am producing those, and by now you will have read several of them. This is simply what I think after a week of turning the results over, and what I have not been able to put down since.
What I cannot put down is not the decline.
The decline is real and it is serious. Reading across the OECD has fallen by something close to a year and a half of learning since 2015. Mathematics sits at the lowest level PISA has ever recorded. One in five fifteen year olds across OECD countries is now below baseline proficiency in science, reading and mathematics at the same time, which was one in six only three years ago. Nobody should wave any of that away, and I am not going to.
But our response to a PISA release is always the same, and it has been the same again. We rank the countries. We admire the systems at the top, we write commentary about the ones that slipped, ministers give interviews, and somebody produces a graph of the last twenty years. Within a fortnight the conversation has finished and we go back to work until the next cycle.
In the middle of all that, there is a question I almost never hear anyone ask.
Who are the children behind these numbers, and does anybody actually know how they got there?
There is a figure in this year's results that I think deserves far more attention than it is going to receive. Socio-economic background explains about twelve percent of the variation in science performance between students within a country.
Twelve percent.
Take two fifteen year olds in the same country. One outperforms the other. The material circumstances they were born into account for roughly an eighth of the distance between them.
I want to handle that finding carefully, because it can be misused in both directions. It does not mean disadvantage has no effect. Advantaged students still outscored their disadvantaged classmates by a wide margin in every system, and the gap narrowed this cycle mainly because advantaged students did worse rather than because disadvantaged students did better, which is nobody's idea of progress.
But it does mean something that I think we skate over. The overwhelming majority of the difference between one child and another is made of something else entirely.
So what is the something else?
That is the question I would want answered if I were running a school this month, and I would want it answered about my own students rather than about my country. It is a far more useful question than where anyone placed in a table.
The something else is not mysterious. Anyone who has worked in a school can name most of it without thinking very hard.
What motivates this child, and whether it is still the same thing that motivated her last year. What she does when the work gets hard, whether she asks for help or goes quiet, because those two children need completely different responses and look identical on a mark sheet. Where she learns well and where she does not, which is partly about the room and largely about who else is in it. How she reasons, and whether the way we happen to teach suits the way she happens to think. Her temperament. Whether she feels she belongs in the building at all, which PISA does ask about and which very few schools ask about their own students. What was going on at home during the term her attendance quietly slipped. Whether the activity she loved in Year 7 is the one she stopped turning up to in Year 9, and whether a single adult noticed.
None of those things appeared in her score.
All of them produced it.
There is a smaller finding in this year's results that makes the same point from another angle. PISA added a new domain this cycle, computational problem solving, and across OECD countries around two thirds of students reached the proficiency level that matters. Only about half reached it while also clearing baseline in science, reading and mathematics. Read that slowly. There is a sizeable group of fifteen year olds who can model a problem and build something that works, and who are sitting below the line in three other subjects. A single mean score for that child, or for the country she lives in, tells you very little about what she can actually do.
Here is the part I find genuinely hopeful, and it is the reason I wanted to write this at all rather than just read the reports and move on.
There is a substantial body of research on early warning systems in schools, built on tracking attendance, behaviour and course performance from as early as Year 6 in order to identify children drifting off track. The finding that has stayed with me is this. Student demographic characteristics turned out not to be powerful predictors. Attendance, course performance and classroom behaviour were. And the researchers make a point of saying why that matters, which is that those are precisely the things a school can act on.
Put that alongside the twelve percent and a rather different picture emerges from the one the headlines are painting this month.
The part of a child's outcome that we tend to treat as fixed by background is smaller than the fatalistic reading suggests. The part that responds to what a school notices, and to what it does about what it notices, is larger.
I find that encouraging. I also think it places a considerably heavier responsibility on schools than the ranking conversation does, which may be exactly why it is the less popular framing.
PISA itself makes the same point, though it puts it differently. Every single participating system has academically resilient students, children from disadvantaged backgrounds who perform highly anyway. PISA can prove that they exist. What PISA cannot do, by design, is tell you what made any one of them resilient, because it is a sample of a population rather than a record of a person.
Only a school can answer that. And most schools cannot, not because they are careless, but because the answer is scattered.
I spent twenty years working in schools before I started building anything for them, and I want to describe something I suspect will be familiar to anyone who has held a leadership role.
Almost every school already holds most of what it would need to answer these questions. It simply holds it in pieces, in places that do not speak to one another.
There is an admissions file containing an interview, an assessment, and a conversation with parents who understand their own child better than any instrument ever will. There is an attendance record. There is a pastoral note written two years ago by somebody who has since moved to another country. There is a list of who signed up for which activity and, far more revealing, who stopped going. There is a teacher who understood exactly why a boy went quiet in the spring term and who mentioned it once to a colleague in the corridor.
None of that information is missing. All of it is unreachable.
So every September the school begins again from a set of grades, and by March has painfully rebuilt an understanding it already possessed eighteen months earlier, and in July it starts the cycle over.
This is where I think the real cost sits, and it is not the cost we usually talk about.
The cost is not that schools fail to notice things. Schools notice a great deal. Teachers are extraordinarily perceptive about children, and they notice far more than they are ever able to record. The cost is that noticing in Year 7 does not reach Year 10.
A concern raised early, by someone who was right, dissolves into a system that had nowhere to put it. Four years later a different teacher meets the same pattern for what she believes is the first time, and begins from the beginning. Multiply that by a few hundred children and you have the accumulated forgetting that sits underneath a great many disappointing results, and none of it will ever appear in an inspection report.
What I have come to believe, and what I now spend my working life on, is that schools need a layer of understanding that sits alongside academic results rather than underneath them.
Not more testing. Schools have plenty of testing, and adding another instrument to a fifteen year old's year is not an act of kindness. I mean a structured way of holding what a school already learns about a child, from the first enquiry onward, so that it accumulates rather than evaporates.
In practice that means being deliberate about the dimensions of a young person that a grade does not capture. Their academic profile, in the sense of how they work rather than what they scored. Their reasoning and aptitudes. Their personality and temperament. Their social and emotional adaptability, which is a clinical way of describing how a child copes when things change. What motivates them and what genuinely interests them. The conditions in which they learn well. How well the environment they are in actually suits them. And then the pattern across all of those, which is usually where the useful insight lives, because no single dimension explains a child and the combination frequently does.
At NovaEd we have built that as eight dimensions, and I am not going to spend the rest of this piece describing our platform. The specific implementation matters far less than the idea, and the idea is available to any school willing to take it seriously, with or without us. Decide what you want to understand about a child. Capture it once. Make it reachable by the people teaching that child. Carry it forward when they move up a year. Treat it as seriously as you treat the mark book.
I want to be precise about what I am not claiming.
I am not claiming that software raises PISA scores. Nobody can claim that honestly, the causal distance is enormous, and any vendor who tells a head of school otherwise should be shown out of the building.
What I do believe, and what the evidence supports link by link, is narrower and more defensible. Understanding a child better makes earlier and better targeted support possible. Earlier, better targeted support is associated with better engagement and better outcomes across a range of studies. Relationship quality between a teacher and a student is associated with engagement and with achievement. A student's sense of belonging shows a small but consistent association with achievement, and recent work suggests the effect genuinely runs from belonging toward attainment rather than the other way round.
None of those effects is individually dramatic. They compound rather than transform. And every one of them measures a school doing something after it noticed something, which is the part people skip. Understanding that nobody acts on changes nothing at all. The data is not the intervention. It only makes the intervention possible, and earlier.
That is the whole of my claim, and I think it is enough.
A PISA score is a photograph of one morning in a fifteen year old's life. It is a useful photograph, taken carefully, and I am glad we take it. But the child who sat down that morning had been assembled over fifteen years by a thousand things the test never asked about, and by a school that either understood those things or did not.
The systems that recover over the next decade will not be the ones that ran the hardest at the test. They will be the ones whose schools could answer, for any child in the building, a question that is much harder than it sounds.
Not what did she score.
Why.