The problem
The Public Accounts Committee reported in March 2026 that "the Department does not know whether home to school transport is achieving value for money", and that DfE lacks the data to oversee the service — it only began collecting who receives transport in February 2025, achieving a 75% response rate with partial answers.
The spend is not marginal. Home-to-school transport cost £2.6bn in 2024-25, of which £2.0bn was SEND transport. SEND transport spending rose 106% in real terms between 2015-16 and 2023-24. The median annual cost is £8,116 per SEND pupil against £1,526 for mainstream.
The underlying driver is demand nobody forecasts well. EHCPs reached 718,838 in January 2026, up 12.5% in a single year — the largest annual increase since plans were introduced in 2014. Only 46.1% are issued within the statutory 20 weeks. Councils lose 99% of SEND tribunal appeals while spending £153m a year defending them.
Government has just written off 90% of historic DSG deficits, worth over £5bn, conditional on councils submitting an approved local SEND reform plan. Those plans are due now, and most councils lack the analytical basis to write a credible one. New deficits continue to accrue on top.
The system
Compass does two connected jobs. It forecasts SEND demand at local authority and settlement level, and it optimises the transport network that demand generates.
The forecasting side combines births, migration, school census, EHCP flow rates and housing completions to project EHCP volumes and specialist placement need three to seven years ahead — the planning horizon for creating specialist places, which is the only durable way to reduce both transport cost and independent placement cost.
The transport side treats school transport as the routing problem it actually is. Given pupil locations, placement locations, statutory eligibility and individual needs, it identifies achievable route consolidation, exposes the cost of each placement decision at the point the decision is made, and quantifies the transport consequence of opening a specialist unit in one location versus another.
The connection between the two is the point. A placement made without visibility of its seven-year transport cost is how a council ends up paying five figures a year to move one child, and no current system shows that number at the moment it matters.
Worked examples
Two situations this system answers
A county forecasts falling pupil numbers and closes a specialist unit. Two years later demand in one district has risen sharply and children are placed forty miles away at roughly £65,000 a place.
Compass forecasts neighbourhood by neighbourhood instead of county-wide, so the rising district shows up even while the county falls. The National Audit Office found this mismatch in two-thirds of councils.
A council must submit a SEND reform plan to unlock its share of a £5bn deficit write-off. It cannot evidence where demand will sit in three years, so the plan is assertion rather than analysis.
Compass produces the forecast the plan needs — how many children, needing what, and where — using only published figures. No individual children’s records are touched.
Data foundation
Every dataset below is open, or its access constraint is stated
| Dataset | Publisher | What it provides |
|---|---|---|
| EHCP statistics | DfE | Plans in force, new plans, assessment requests, 20-week timeliness, by local authority. Published annually. |
| School census / Get Information about Schools | DfE | Pupil-level characteristics, SEN status, school capacity and location. |
| SCAP school capacity survey | DfE | Place planning capacity by authority and phase. |
| ONS births and population projections | ONS | Cohort sizes feeding the forecast, by small area. |
| Housing completions and permissions | MHCLG | New development driving future pupil numbers at settlement level. |
| Indices of Multiple Deprivation | MHCLG | Correlates strongly with SEN identification rates. |
| OS Open Roads / journey routing | Ordnance Survey | Network for genuine travel-time calculation rather than straight-line distance. |
| SEND Tribunal statistics | MoJ | Appeal volumes and outcomes, for measuring decision quality. |
Capabilities
Benefits
For government
- Answers the PAC finding directly, giving DfE the oversight data Parliament has said it does not have.
- Supports the condition attached to the £5bn deficit write-off — councils must submit credible SEND reform plans, and most currently cannot evidence them.
- Addresses the fastest-growing local government cost line with a forecasting horizon long enough to actually change it.
- Provides MHCLG with evidence on whether the new home-to-school transport funding formula works — which PAC noted it has no plan to monitor.
For the public
- Children spend less time in taxis. Some SEND pupils currently travel hours a day.
- Faster, better-evidenced placement decisions reduce the adversarial tribunal process families are forced into — 25% of families reduce working hours fighting for support.
- Specialist places created where children actually live, rather than wherever capacity happened to exist.
Delivery
Phasing
Risks & mitigations
Route consolidation must never override an individual child’s assessed needs. Needs constraints are hard limits in the model, not weighted factors.
EHCP growth has repeatedly outrun projections. Published back-testing against outturn, with intervals wide enough to be honest.
Children’s data of the highest sensitivity. Operates on aggregate and pseudonymised data by default; record-level work only under the authority’s controllership.
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.