The Four-Tier System: When Support Comes Too Late

A system built on judgment, not understanding

Before we talk about outcomes, we need to talk about how this system is being proposed and discussed, and how it would operate in real schools, with real children.

The proposed four-tier SEND system is often described as a way to “structure” support. In practice, it would structure delay. Each tier would require a judgment, a period of waiting and evidence of failure before movement is allowed. That may sound orderly on paper, but children do not develop on paper. They develop in real time.

Under this model, the first decision about a child’s “level of need” would rarely be made by a specialist. It would usually begin with a classroom teacher. Teachers are skilled professionals, but most are not trained in neurodiversity, auditory or visual processing differences, sensory integration, or executive functioning. This is not a criticism of teachers; it is a reality of the training system they work within. Yet a tiered model would place the responsibility for early judgment on staff who are not equipped to identify complex and often invisible needs.

Many neurodivergent differences do not present as obvious learning difficulties. A child who misses instructions may be struggling with auditory processing, not listening. A child who rushes work may be overwhelmed, not careless. A child who reacts emotionally may be dysregulated, not defiant. Without specific training, these differences are easily misinterpreted. Once that framing sets in, any future escalation of support becomes slower and harder.

When support is tied to time rather than need

A four-tier approach assumes that incorrect judgments will be corrected over time. But time is exactly what children do not have.

Support at the lower tiers would be informal, time-limited, and not legally protected. It could be withdrawn, reset, or replaced, each time requiring the child to demonstrate once again that they are struggling. Neurodivergence, however, is not a short-term issue. When support ends, and difficulties resurface, the child risks being seen as failing to respond, rather than as needing sustained support.

Inconsistency would be built into the system. Movement between tiers would depend on the confidence of individual teachers, the capacity of SENCOs, school budgets, and local authority thresholds. Two children with identical needs could receive very different support depending on where they live. This is not a needs-led approach; it is a resource-led one.

The children this model would miss

The most serious risk emerges when difficulties appear later.

Many children cope throughout primary school with structure, routine, and adult scaffolding. It is often in secondary school, when demands increase sharply, that difficulties escalate. Organisation, independent learning, emotional regulation, and sensory tolerance suddenly become critical. For some children, particularly those with ADHD, autism, or both, needs intensify rapidly at 12, 13, or 14.

A tiered system is not designed for rapid escalation. It would require observation, evidence, interventions, and time. While this process unfolds, children would fall behind academically, lose confidence, develop anxiety, and begin to disengage from school. Behaviour would become the signal not because the child is difficult, but because they are overwhelmed. By the time statutory support is considered, many would no longer be simply struggling — they would be burnt out.

The predictable outcome

This process would not take months. It would take years.

Some children would never reach the highest tier at all. They would not improve; they would disengage. They would move onto part-time timetables, be informally excluded, or disappear from education altogether. At that point, the system would describe them as too complex, too anxious, or too disengaged — without acknowledging that delay was built into the design.

The four-tier system is often presented in reform discussions as a form of early intervention. But early intervention does not wait for repeated failure. It does not rely on misinterpretation being corrected eventually. And it does not accept long-term harm as a necessary step toward support.

While this model has not yet been formally introduced, its assumptions are already shaping how support is discussed and rationed. If implemented without fundamental change, the outcome is predictable.

Support would come too late — after the damage is already done.