The problem
The numbers look contradictory until you look below local authority level. England has 1.2 million unfilled school places — 13% of all capacity, including 680,000 unfilled primary places, the highest since collection began. At the same time about 3,000 schools are at or over capacity.
The National Audit Office explained why in April 2026: "Of local authorities forecasting falling primary school pupil numbers, 66% expect to see numbers increase in one or more of the smaller areas for which they gather data." Surplus and shortage coexist inside single councils. It is a spatial allocation failure, invisible in every published statistic.
The unit where it happens is the pupil planning area. There are 3,651 of them across 153 authorities, each defined by its own council. DfE publishes school capacity data at planning-area level — and the boundaries are published nowhere as open geospatial data. No shapefile, no GeoJSON, no lookup to any standard geography. Councils receive rows they cannot map, compare, or join to anything.
The inputs DfE expects councils to forecast from do not exist at that geography either. ONS publishes 2025 fertility at country level only. There is no national site-level housing completions dataset. And DfE’s own National Pupil Projections are national level only, with the horizon cut from ten years to five because of migration uncertainty.
The consequence is measurable: the London Assembly had to send freedom of information requests to all 33 boroughs to count how many schools had closed and why, because the data does not exist. Meanwhile EHCPs reached 718,838 with 211,400 pupils in special schools against 160,000 special school places — and the NAO records that DfE "does not know how many spaces are available in mainstream schools or other settings."
The system
Catchment starts with the cheapest high-value fix in this entire portfolio: publishing an open, versioned geography for the 3,651 pupil planning areas, reconstructed from the school membership DfE already publishes, with a lookup to standard census geographies. One artefact, and every join in the domain becomes possible.
On that foundation it builds small-area demand forecasting — apportioning births, internal migration and new-build completions to planning areas to project reception intake where it actually occurs, rather than at the authority level where the signal cancels out.
It then makes the NAO’s finding visible: a national map of where surplus and shortage sit within the same authority, authority by authority and area by area.
And it offers the one genuinely constructive answer to the surplus problem. Rather than closing schools that will be needed again when housing lands, it identifies mainstream schools with persistent surplus within travel distance of unmet specialist demand — converting empty classrooms into the scarcest asset in the system. DfE’s own estates strategy commits to re-using surplus space and has no tooling to find it.
Worked examples
Two situations this system answers
A council with falling rolls closes a primary school. Two miles away another is over capacity and turning children away. Both facts are true, and neither is visible in council-level statistics.
Catchment publishes the missing boundaries for all 3,651 areas, so empty seats and overcrowding show up side by side inside the same council.
Basic Need capital is allocated at planning area and year group with no offsetting between areas — so £1.096bn is distributed on a geography that is published nowhere and cannot be mapped.
Catchment rebuilds those boundaries from a list the department already publishes, and gives them away free.
Data foundation
Every dataset below is open, or its access constraint is stated
| Dataset | Publisher | What it provides |
|---|---|---|
| School Capacity (SCAP) | DfE | Capacity, forecasts and planned changes at national, regional, authority, planning area and school level. Notably absent from the DfE statistics API. |
| Get Information About Schools (GIAS) | DfE | Daily bulk CSV, no authentication. 52,486 establishments, 135 fields, including closure dates and reasons. |
| National Pupil Projections | DfE | National level only — the gap Catchment fills locally. |
| EHCP statistics | DfE | Plans by placement type, for the specialist capacity gap analysis. |
| ONS births and internal migration | ONS | Cohort inputs, apportioned to planning areas. |
| EPC new-build lodgements | MHCLG | Address-level, near-real-time completion proxy — the only one available nationally. |
| ONS Open Geography Portal | ONS | Standard boundaries for the planning-area lookup. |
Capabilities
Benefits
For government
- Answers the NAO’s April 2026 findings directly, including that DfE does not use capacity data to monitor how schools respond.
- Gives the DfE Pupil Place Planning team — the named owner of SCAP and Basic Need — the sub-authority evidence its own allocation formula depends on.
- Supports the February 2026 ten-year estates strategy commitment to re-use surplus space, which currently has no supporting tooling.
- Connects mainstream surplus to the specialist shortage driving high needs deficits, linking two problems currently managed on entirely separate tracks.
For the public
- Fewer schools closed that are needed again five years later, and fewer children in overcrowded classrooms two miles from empty ones.
- Specialist places created by converting surplus capacity, reducing both waiting times and transport distances for children with EHCPs.
- Parents and governors can see the actual demand picture for their area rather than an authority-wide average that hides it.
Delivery
Phasing
Risks & mitigations
Reconstructed geographies are approximations of authority-defined areas. Published with explicit confidence and a correction route for councils to submit their true boundaries — which improves the asset over time.
School closure is intensely political. Catchment provides evidence and explicitly models reversibility rather than recommending closures.
GIAS warns its bulk fields and field order can change without notice. Handled by schema-tolerant ingestion with change alerting.
Sources
All sources checked in August 2026. Figures carry the reference period of their source, which may differ from publication date. Where a figure could not be verified against a primary source it is not used.