A child can be visible to five public services and still have nobody with enough time to help.

This is the problem at the centre of the government’s new data-sharing plan.

On 5 June 2026, ministers announced a project to connect information from health visiting, education, and childcare providers. The stated aim is to help more children arrive at school ready to learn.

The public case is straightforward. A health visitor may notice one problem. A nursery may notice another. Months later, a school may be the first service to see the wider pattern. When records sit in separate systems, or on paper, the family can end up carrying the whole burden.

Parents repeat the same story. Professionals miss details. Support arrives late.

The government says almost a third of children start school without the expected basic skills. Among children eligible for free school meals, the figure is almost half.

So the proposal has an obvious appeal. The real question is what follows.

Seeing the child is only the first step

A public service can know a child needs help and still fail to provide it. Data language often hides that gap.

A shared record does not create a speech and language therapist. A dashboard does not give a teacher more time. A linked system does not repair poverty, housing stress, transport problems, disability barriers, trauma, food insecurity, or long waiting lists.

Earlier identification may help. Some children are missed because services are badly joined up. But visibility is not care.

Care begins when the need is met.

If a child is marked as needing speech and language help, does the help arrive? When a family is struggling with attendance, does anyone ask about transport, illness, anxiety, housing, or unpaid caring? After a nursery flags developmental delay, does a specialist see the child soon enough to matter?

These are the tests. A system that only records unmet need has not fixed the problem. It has made the failure easier to count.

This data-sharing plan sits beside a wider set of SEND reforms.

The government is also promoting “Experts at Hand”, a programme meant to bring speech and language therapists, occupational therapists, educational psychologists, specialist teachers, and other professionals into mainstream settings.

That could be useful.

Many families already know the pattern: a child struggles, the school sees it, the parent asks for help, and then everyone waits. Forms are filled in. The same story is explained again. The right person may eventually appear, but often too late.

Earlier specialist support would be better than a long paper chase.

The data project and the SEND support plan have to be judged together. If data-sharing helps bring support into the room, it may help children. If it mainly gives the system a cleaner way to record delay, it becomes another layer of administration.

The public should not be impressed by joined-up records unless they lead to joined-up help.

Labels can follow a child

Records do not only store facts. They create categories.

One file may describe a child as “not school ready”, a parent as “non-engaging”, a family as “hard to reach”, or a household as “high risk”. Those phrases may begin as shorthand. Later, they can harden into reputation.

A soft judgement written in one service can shape how another service sees the family. A teacher, health visitor, social worker, or council officer may meet the record before they meet the parent. The family then has to answer a version of themselves created somewhere else.

Some labels are necessary. Safeguarding records matter. Developmental concerns matter. Professionals should not be forced to forget important information. A fair system still has to separate evidence from impression.

“Child missed eight sessions” is a checkable fact.

“Parent seems disengaged” is a judgement.

“Family did not attend because the bus route changed after a housing move” tells a different story again.

Those differences matter. If the system flattens them into one risk label, the family loses the chance to be understood.

Families need a right of reply

Unless there is a clear safeguarding reason not to, parents and young people should be able to see what is being said about them.

They should know who wrote a concern, what evidence supports it, who has seen it, and how it may be used. Wrong, incomplete, or unfair records need a clear challenge route.

That challenge should sit inside the record. A complaint form outside the system is not enough. By the time a complaint is answered, the label may already have travelled.

A fairer system would let a parent add context:

“The absence was caused by temporary housing.”

“The appointment was missed because the letter went to the old address.”

“The behaviour concern followed sensory overload.”

“The family asked for support before this became a safeguarding issue.”

Context does not remove professional judgement. It makes the record less lazy.

Support records should not become punishment records

There should be a firewall between support and punishment.

A record created to get a child help should not quietly become evidence against the child or family. Speech and language concerns should lead to speech and language support. Attendance concerns should lead to practical questions before enforcement. Developmental concerns should not become a general mark against the child.

Safeguarding is different. Serious concerns must be shared when a child may be at risk. Even there, the system needs discipline. It should record what happened, who said what, what is known, what is suspected, and what remains uncertain.

Poor wording can do real damage. A vague concern can become a file note. The file note can become a pattern. The pattern can become a decision.

Once that happens, the family may never know which sentence changed the way professionals looked at them.

The youth justice warning

The same public-service logic is appearing in youth justice.

The Ministry of Justice has created an advisory panel on preventative analytics for youth justice. Its work involves machine learning, artificial intelligence, and advanced analytics. The stated aim is early intervention: finding where support may be useful before harm grows.

The aim sounds reasonable. The warning is in the method.

Prediction systems can turn social need into risk. Poverty, disability, family stress, school exclusion, contact with services, or previous safeguarding involvement may start to look like signs of future offending.

A child who needs help can become a child being watched.

The government says this work needs safeguards. That is the right word. But safeguards have to be built into the machinery, not added as a paragraph after the policy is already moving.

The hardest question is simple: when the state predicts risk, does it send help or suspicion?

The safeguards should be visible

Children’s data needs a higher test than ordinary admin. The Information Commissioner’s Office says children’s personal data needs extra care. Sharing should have a compelling reason, and the best interests of the child should be a primary consideration.

That principle should be visible in the design of the system. A proper child data system would show:

  • why each piece of information is being shared
  • who can see it
  • how long it will stay there
  • whether it is fact, judgement, allegation, or family context
  • how the family can challenge or correct it
  • what support becomes available because of it

The last point is the most important. If no support follows, the data-sharing case becomes weak.

A child does not need a better file instead of help.

The TWIS test

This is not a simple argument against data-sharing. Some information should be shared. Disconnected services can hurt children. Professionals need enough information to act safely and quickly.

The problem starts when the state treats visibility as support.

A child can be tracked across systems while still waiting for help. A family can be flagged, scored, discussed, and referred without anyone fixing the practical barrier in front of them.

That is not reform. It is managed delay.

Before any child data project is praised, five questions need clear answers.

What support becomes available because of the data?

Can the family see and challenge what is recorded?

Does the system separate fact from judgement?

Are old labels reviewed or removed?

Is this backed by staff, time, and services?

If the answers are unclear, the reform is not ready to be trusted.

The aim should not be to make children easier to track. The aim should be to make help easier to reach.