Ofsted’s similar schools model: what it changes for multi-academy trusts
Purple Education
9 min read
From September 2026, Ofsted stopped judging achievement mainly against the national average. Every mainstream primary and secondary in England now has a list of 50 statistically similar schools, and that comparison sits inside the inspection data summary report that inspectors read before they walk through the door.
For a standalone school, that is a change of reference point. For a multi-academy trust it is something more awkward. The benchmark most central teams use — school against school, school against the trust average — is no longer the benchmark the inspector is holding.
What the similar schools model actually is
The Department for Education and Ofsted built the model using regression analysis to work out which contextual factors most strongly predict differences in attainment. Four groups of variables feed it:
Prior attainment — key stage 1 results for the key stage 2 comparison, combined key stage 2 reading and maths for the key stage 4 comparison
Disadvantage — pupil premium eligibility, IDACI (neighbourhood income deprivation) and POLAR4 (progression to higher education)
Cohort — school size and pupil stability
Inclusion — proportions of pupils on SEN support, with EHC plans, and with English as an additional language
The output is a list of 50 similar schools for each mainstream primary and secondary. They will not be identical on every measure, and Ofsted is explicit that the data is not used to make grading decisions. It is there to help inspectors understand context and to shape the conversation about achievement.
Two things follow from that. The first is that "we're above the national average" no longer closes the discussion. The second is that a trust's schools are now each being read against a group they are not in — a group chosen by statistical similarity, not by geography, sponsor or shared improvement strategy.
Why this lands hardest on achievement
Achievement was already the grade schools were most likely to fall short on.
Of the 921 schools inspected between 10 November 2025 and 30 April 2026, 70 per cent were judged at the expected standard or better for achievement. Twenty-seven per cent needed attention and 3 per cent required urgent improvement. Compare that with personal development and wellbeing, where 95 per cent met the expected standard or better.
The gap widens in schools serving more disadvantaged communities. Among schools with above-average free school meals eligibility, 57 per cent reached the expected standard or better for achievement, and 36 per cent were told the area needed attention.
One figure from Ofsted's own statistical commentary deserves more attention than it has had. Of 126 high-disadvantage schools graded needs attention or urgent improvement on achievement, 54 per cent were nonetheless judged to be at the expected standard or better for leadership and governance. Leadership was not the problem. The outcomes data was.
Ofsted also published the attainment profiles sitting behind each achievement grade. For disadvantaged pupils at secondary, average Attainment 8 was 35.9 in schools graded at the expected standard, against a national figure of 34.9 for disadvantaged pupils; 47.2 in schools graded strong; and 63.8 in schools graded exceptional. At primary, the equivalent figures for disadvantaged pupils reaching the expected standard were 52 per cent, 65 per cent and 86 per cent, against a national figure of 47 per cent.
Those are averages, not thresholds, and Ofsted's reading of them is that inspectors are allowing for context without normalising lower achievement for disadvantaged pupils. But they give trust leaders something they did not have a year ago: a rough sense of what the data behind each grade tends to look like.
Three things this changes for trust central teams
Internal rankings stop being the answer. A school that sits at the top of your internal attainment table may be mid-table against its 50 comparators, and one you have been worrying about may be doing better than you thought. Because the comparator groups are matched on context, a trust with genuinely mixed intakes can no longer hold every school to a single internal bar and expect that to line up with how each one is judged.
Improvement narratives need a different axis. "Up four points on last year" is a weaker line if the comparator group moved six. Progress over time is now read relative to schools facing similar circumstances, which means the trust-level improvement story has to be told school by school, against each school's own group.
Disadvantage is measured twice. Inspectors look at whether disadvantaged pupils attain in line with disadvantaged pupils nationally, and separately at how the school performs against similar schools. Those are two different questions, and a trust can answer one well and the other badly.
What it does not mean
It is worth not over-rotating on this. The similar schools data is contextual information in the inspection data summary report. It is not a grade, not a threshold, and not a target.
The model is also contested, and trust leaders should know the arguments. Location is not one of its variables, which Schools North East, ASCL and former trust CEO Jonny Uttley have all criticised — coastal and rural isolation, and the specialist capacity within reach of a school, simply are not represented. Professor Georgie Leckie has pointed out that ethnicity was excluded despite research showing it to be the third most important factor after prior attainment and free school meals eligibility. Permanent exclusion rates are not accounted for either, so a school excluding fifty pupils a year and a school excluding none can appear in the same group. The DfE says it will continue to improve the model.
None of that makes the comparison safe to ignore. It does mean that a trust with coastal, rural or highly mobile intakes has a legitimate case to set out, and is better off making it deliberately in board papers than discovering it mid-inspection.
It is also worth reading this alongside what is coming for trusts themselves. Academy trust inspections are expected from 2027–28, and the evidence base that satisfies a school inspection is largely the same one a trust inspection will draw on.
Attendance and inclusion are moving the same way
The September 2026 toolkit updates point in the same direction. Inspectors now weigh attendance with SEND context in mind, looking specifically at the impact of a school's work on attendance for pupils with SEND, and at how leaders work with parents and local authorities where pupils hold EHC plans. Schools are expected to publish an inclusion strategy setting out how funding will meet cohort needs, and Ofsted has introduced the phrase "leaders who have an inclusion role" to signal that inclusion is not the SENDCO's job alone.
The attendance backdrop makes that harder. 2025/26 was the first post-pandemic year with no meaningful reduction in absence. Persistent absence ran at 23 to 27 per cent across regions at secondary and 14 to 16 per cent at primary. In 2018/19 those figures were 14 per cent and 8 per cent.
The plateau matters because it removes a line a lot of trusts have been using. "We're improving" no longer stands on its own when the sector as a whole has stopped improving. What replaces it is a more specific claim: this cohort, this intervention, this shift.
Five questions worth being able to answer this term
For each school, how does attainment compare with its similar schools group — not the trust average, and not the national average?
How do disadvantaged pupils in each school compare with disadvantaged pupils nationally?
Which pupils are on a trajectory towards persistent absence right now, rather than which pupils were persistently absent last year?
Where do SEND, disadvantage and low attendance overlap in each school, and what is in place for those pupils?
For any intervention running across the trust, what is the evidence that it worked?
Most trusts can answer some of these. Very few can answer all five across every school, in the same view, in the same week the question gets asked. That is usually not an insight problem. It is a plumbing problem — attendance in one system, behaviour in another, assessment in a third, provision tracked in a spreadsheet somebody maintains in their own time.
Purple was built for exactly that gap: bringing attendance, behaviour, achievement, provision, inclusion and pupil voice into one place so trust leaders can see patterns across schools rather than assembling them by hand each term. Purple has recently been accepted onto the Education Endowment Foundation's data collection panel as an attainment testing provider.
Over to you
The question we keep coming back to is whether the similar schools model gives trust leaders something genuinely useful, or just another comparison to manage.
If you lead a trust: has the similar schools data changed how you look at your own schools this term — or is it sitting in the IDSR unread? And for those of you in coastal, rural or high-mobility contexts, does the comparison group feel fair?
We would like to hear what you are seeing. And if the harder problem is that the data itself lives in six different places, that is the part we can help with.