How the School Chooser Score works
No single number can sum up a school. But parents compare schools anyway, so the score should at least be fair, honest about what it can't tell you, and open for anyone to check. This page is the whole method: every figure that goes in, how much each one counts, and the checks every run has to pass before it goes live. Version v1, last run 13 September 2026.
The short version
We take what is published about a school: exam results over three years, the progress pupils make, the latest Ofsted inspection, attendance and behaviour, whether things are getting better or worse over five years, and what parents told Ofsted. We measure each of those against the 50 schools in England whose pupils are most like this school's. For small schools, where one unlucky year can swing the numbers, we pull the figures part of the way towards that group's average. Then we put the pieces together into one score out of 100, where 50 is the middle school in the country. Every figure has a margin of error, so the score does too, and we show it rather than hide it.
Who is scored
Mainstream state-funded schools in England with a primary (key stage 2) or secondary (key stage 4) phase. All-through and middle schools get a score for each phase. We do not score special schools, alternative provision, nursery schools, sixth-form colleges or independent schools: their results aren't published in a way that can be compared fairly, and it is better to say so than to pretend. Their pages still show everything that is published about them. A school also goes unscored when it has neither results nor an inspection, or fewer than two components in total.
What goes in, and how much each part counts
| Component | Primary | Secondary | What goes in |
|---|---|---|---|
| Attainment | 25 | 20 | Primary: % reaching the expected standard in reading, writing and maths (weight 0.4), % at the higher standard (0.2), average reading scaled score (0.2), average maths scaled score (0.2). Secondary: Attainment 8 (0.4), % grade 5+ in English and maths (0.25), EBacc average point score (0.2), % grade 4+ in English and maths (0.15). Three years, weighted 0.5 / 0.3 / 0.2 and by cohort size. |
| Progress | 20 | 20 | The Department for Education's own value-added measure where it is fresh: Progress 8 at secondary, the average of reading, writing and maths progress at primary. DfE did not calculate Progress 8 for the 2025 or 2026 cohorts and paused primary progress for 2025, so when the latest measure is more than three years old we substitute our contextualised attainment (below) and say so on the page. |
| Ofsted | 20 | 20 | The latest full inspection, converted to a score from 0 to 100. Report cards (November 2025 framework): each area on the five-point scale scores 100 / 82 / 62 / 30 / 5, weighted leadership 1.5, achievement 1.5, inclusion 1.25, curriculum 1.25, attendance and behaviour 1, personal development 1, early years 0.5, post-16 0.5. Legacy grades: outstanding 95, good 70, requires improvement 35, inadequate 5; inspections from September 2024 to November 2025 with no overall grade use the weighted sub-judgements. Safeguarding not met caps the score at 30; a category of concern caps it at 20. An ungraded visit that raised concerns subtracts 10. Ofsted's open data only carries graded inspections from September 2019, so a school whose last graded inspection was earlier and has since had an ungraded visit confirming it “remains good” or “remains outstanding” is scored on that confirmed grade, dated at the visit; about a quarter of schools are in this position. The weight decays with age: full for inspections up to three years old, falling to 40% at eight years, with the freed weight going to the data components. |
| Context | 10 | 10 | How far the school's attainment sits above or below what its intake would predict: the residual of a ridge regression of attainment on free school meal eligibility (ever 6), English as an additional language, EHC plans, SEN support, neighbourhood deprivation (IDACI quintile or IoD 2025 decile), school size, prior attainment mix (secondary), selection and single-sex status. |
| Attendance and behaviour | 10 | 10 | Overall absence (0.5) and persistent absence (0.5) in the latest year; secondaries add suspensions per 100 pupils (0.3). Lower is better. |
| Trend | 10 | 10 | The slope of the school's yearly attainment (0.7) and absence (0.3) relative to national over the last five years. "Improving" and "declining" labels use a threshold of ±0.15 standard deviations a year. |
| Parent View | 5 | 5 | The share of parents who would recommend the school, from Ofsted Parent View where at least 10 parents responded since the November 2025 question change, shrunk toward the national average with a prior weight of 20 responses. |
| Destinations | – | 5 | The share of pupils in sustained education, employment or apprenticeship the year after GCSEs. |
When one part is missing, its share is spread across the others, and the school's page says what was and wasn't used.
Wales, Scotland and Northern Ireland
Wales. Secondary schools are scored on the same 0 to 100 scale, but against other Welsh schools only, never against English ones. Wales publishes different measures (the Capped 9 points score with literacy, numeracy and science point scores, from the Welsh Government's My Local School), Estyn has published no grades since 2022, and there is no value-added measure or Parent View. So the Welsh secondary score rests on results over three years (40%), results given the intake (25%), attendance (20%) and direction of travel (15%). Wales publishes no results for primary schools, so Welsh primaries have a profile without a score.
Scotland. Scotland publishes no school-level exam results or inspection grades as open data, so Scottish schools have a profile only: roll, free school meals, additional support needs, staffing, class sizes and the deprivation mix of their pupils, from the annual pupil and teacher census.
Northern Ireland. The Education and Training Inspectorate publishes no grades as open data and there is no progress measure, so schools in Northern Ireland have a profile rather than a score. Post-primary profiles carry the Department of Education's Year 12 GCSE and Year 14 A level measures, compared with grammar or non-grammar schools as appropriate because grammar schools select by ability; every profile has enrolment, free school meals, special educational needs, newcomer pupils and attendance from the October census and the attendance tables.
How schools are compared
- Every indicator is turned into a z-score within its year and phase (how many standard deviations from the mean of scored schools) and clipped at ±3.
- Each school's 50 most similar schools are found by Mahalanobis distance on the standardised intake covariates above. Grammar schools are only compared with grammar schools.
- Each data component is shrunk toward the mean of those 50 schools using an empirical-Bayes weight of n / (n + k), where n is the latest cohort and k is 25 for primaries and 60 for secondaries. A primary with a cohort of 12 therefore keeps only a third of its own signal; one with 120 keeps more than 80%.
- The composite is the weighted mean of the shrunk components. The score is its percentile among scored schools of the same phase, so 50 is the median school and 90 beats nine schools in ten.
The likely range and the bands
Each component carries a standard error: from the published confidence interval for progress measures, from cohort size for percentages (roughly 3 / √n for a single year, less once three years are pooled), and fixed allowances for inspection (0.3), trend (0.5) and parent views. These propagate through the weights to the composite, and the likely range shown on every page is the 90% interval mapped onto the percentile scale. Bands are: well above similar schools (score 85+), above (65 to 84), in line (35 to 64), below (15 to 34), well below (under 15). When less than 60% of the likely range falls in the school's band we mark it borderline and name the neighbouring band. Rankings show the range next to every score for the same reason: two schools whose ranges overlap should not be read as one beating the other.
Checks every run must pass
- Band shares are close to 15 / 20 / 30 / 20 / 15 in each phase (they are, by construction).
- The score is less correlated with free school meal eligibility than raw attainment is. In the current run the correlation with FSM is -0.37 for primary scores against -0.40 for raw attainment, and -0.49 against -0.60 for secondary. Contextualisation is working, but not perfectly: schools serving poorer intakes still score lower on average, mostly because Ofsted grades and attendance, which we do not contextualise, also track deprivation.
- No school moves more than two bands between runs without a new inspection or results year.
If a run fails a check it is kept for us to look at but never published.
What the score is not
- It is not a measure of how happy your child will be there, how good the pastoral care is, or whether the head is any good. Those things matter more than anything here.
- It is not the same as Ofsted's judgement, the DfE's league tables, or any other site's stars. Each answers a different question.
- Small differences mean nothing. A school scoring 62 and one scoring 58 are, on this evidence, the same.
- Independent schools are not on the same scale at all: DfE's figures for them exclude IGCSEs, so we do not score them.
Sources, dates and changes
All inputs are official open data listed on the data sources page, with the publication each figure came from and when it was last refreshed. The score is recalculated whenever a source changes. Method changes get a new version number and an entry here.
Version history
- v1, September 2026: initial method as described above.
Think a figure is wrong?
Every number on a school page comes from a published file. If one is wrong, or has changed since we loaded it, tell us the school's URN and which figure. We check it against the source and put it right within a few days. Schools can also ask for a note to be added where a published figure is misleading (for example, a merger that split a cohort).