Permanent exclusions fell. Did lost learning simply move inside school?

01 / The national turn

A welcome turn, not an all-clear.

The annual fall came after post-pandemic increases, but the remaining scale is substantial. Some 326,700 pupils had at least one suspension in 2024/25. Among them, 94,400 missed more than one school week cumulatively and 45,600 missed more than two. Primary suspension rates rose even as the overall rate fell.1

The national collection can tell us about formal suspensions and upheld permanent exclusions. It cannot show how long pupils spent in disciplinary removal rooms, what curriculum they received there or how often they returned. Arbor’s publication is useful precisely because it exposes that measurement gap, not because it fills it nationally.12

02 / The Arbor signal

Read the proprietary signal within its boundary.

Arbor analysed records from 13,000 schools using its MIS across four academic years. Its secondary-school incidence series rose steadily – 5%, 7%, 10% and 13% – but the customer base grew during the period. Arbor therefore compares pupil shares rather than raw totals. Representativeness is not established.2

The same dataset reports at least one removal for 22% of disadvantaged secondary pupils, against 10% of non-disadvantaged pupils, and for 18% of pupils with an EHC plan. These figures describe who was recorded at least once. They say nothing about minutes, repeat exposure, work completed or whether definitions were identical between schools.2

03 / Defensible removal

Removal can be legitimate – and still requires scrutiny.

DfE advice defines classroom removal as a serious disciplinary sanction for a limited period. It should be necessary, normally follow other classroom strategies and allow education to continue in a supervised setting. That education may differ from the mainstream curriculum but should remain meaningful; parents should be informed the same day.3

The same guidance expects a suitable, resourced space, trained supervision, a clear reintegration process and internal analysis of repetition and disproportionate effects. Prolonged removal requires explicit headteacher agreement and extensive educational and pastoral support. The governance question is therefore not “removal or no removal”, but whether it is necessary, educational and effective.3

04 / Inequality and response

Inequality needs rates, duration and response.

National disparities remain stark. In 2024/25 the suspension rate was 26.66 for pupils eligible for free school meals, against 5.41 for other pupils. Rates were 28.86 for pupils receiving SEN support, 26.45 for pupils with EHC plans and 6.78 for pupils with no identified SEN.1

Local analysis should add internal-removal incidence and duration without assuming discriminatory intent from a rate difference alone. DfE advice asks schools to investigate protected-characteristic patterns, underlying factors and whether frequently removed pupils need a pastoral review, SENCo investigation or different support.3

05 / The learning ledger

Govern learning lost, not sanction labels.

The statutory exclusions guidance asks governing boards to challenge the whole picture of pupil movement, including permanent exclusion, register removal, off-site direction and other movement. Boards should examine who moves, why, when and whether patterns among pupils with SEND point to weaknesses in support.4

That principle should extend to time away from the usual curriculum. Off-site direction must be temporary, improvement-focused and accompanied by a broad and balanced education. Internal removal should provide meaningful learning. A single ledger can distinguish curriculum displacement from learning that was genuinely not provided.34

Leadership questions

Questions that turn the development into a governing conversation.

School Connection intelligence lens

What the national Observatory can add, and where it must stop.

Exact Observatory test: for each school term, create de-duplicated episode records for suspension, disciplinary classroom removal, off-site direction and authorised reduced timetables. Keep planned sensory, nurture or therapeutic separation separately coded. Removed-from-usual-curriculum minutes = scheduled curriculum minutes affected by those episodes minus overlapping minutes. Unprovided-learning minutes = removed-from-usual-curriculum minutes minus minutes of verified meaningful education delivered and completed.34

Report pupils affected, events, median and total minutes per 100 pupils, repeat exposure at three and five episodes, reintegration completed, and recurrence within 10 and 20 school days. Apply a substitution flag where the formal suspension or permanent-exclusion rate falls while disciplinary internal-removal minutes per 100 pupils rise for two consecutive terms under stable definitions. Label this a signal, never a causal finding. Disaggregate aggregate rates by phase, year, sex, disadvantage, ethnicity and SEN/EHCP status, showing both cohort share and removal share.134

Privacy and small-cell limits: use pseudonymised pupil records only inside authorised local analysis. Publish no pupil, family, reason narrative or placement history. Suppress cells below 10, apply complementary suppression, and combine periods or geographies where linked characteristics could enable re-identification.

What we are monitoring next

Publication is the beginning of the watch.

• The next DfE annual exclusion release, especially phase and subgroup movements.
• Any MIS analysis publishing duration, repeat exposure, coverage and consistent definitions.
• DfE revisions to behaviour, removal or pupil-movement guidance and implementation evidence.