As you may be aware… the public debate is almost exclusively about ownership, profit, private equity (or some combination thereof). The actual failure is upstream and far older.

Last week, the Department for Education published Delivering the children's social care reset, a 45-page implementation plan for local partners covering 2026 to 2029.

It is the most integrated picture we have of where national policy is going after the Children's Wellbeing and Schools Act received Royal Assent in April, backed by historic numbers:

In any other context, a document this concrete would mark a turning point!

The very same day, the BBC's Noel Titheradge, released a piece with FOI evidence of unregistered children's homes costing local authorities up to £2 million per child per year.

Within hours the Children's Commissioner posted a response on LinkedIn calling on Ofsted to "crack down" on the providers as a matter of urgency.

The Reset is real. The cycle is real. They are not yet pointing at the same problem. This article is about the gap between them.

Confirmed cases of unregistered children's homes rose 372 per cent between 2020/21 and 2024/25. On a single day in September 2025, the Children's Commissioner identified 669 children in unregistered placements across 153 local authorities, close to Ofsted's confirmed total for the previous entire year.

The Public Accounts Committee currently puts the figure at around 800, despite the 2021 ban on the practice for under-16s. These figures almost certainly understate the scale: data collection began only in 2020/21, both Ofsted and local authorities acknowledge significant voluntary underreporting, and FOI requests by academic researchers in this space have returned sparse and inconsistent responses.

Every cycle of press coverage names private providers as the primary cause. Every cycle produces calls for profit caps, prosecution, and registration drives.

Marie Tucker spent twenty-five years as a commissioning manager inside local authorities. She writes publicly about how the system actually works, and has two taxonomies that explain why policy aimed at the wrong category keeps failing.

The first is a taxonomy of operators. PE-backed providers are large groups, often well graded, with the scale to deliver prices smaller providers cannot match. Small ethical operators sometimes make more than intended because of how local authorities structure occupancy into their contracts. And a third category. Her words:

"Those who are going: all I need to do is open a children's home and charge £20,000 a week, whether I'm registered or unregistered. Those are the ones that make me angry. Everyone seems to be lumping them all together."

The dominant narrative attributes the crisis primarily to the first category. There are vocal, care-experienced campaigners whose voices carry real weight in this debate, and who locate the problem in decades of marketisation since the 1980s. There is real history behind that argument, and MAP Adventures (the company at the centre of the recent case in the news) was itself a for-profit limited company. It is also worth quickly remembering why the sector exists at all. Children's homes were once run largely by local authorities; the private and voluntary sector grew through the 1990s and 2000s as councils retreated from direct provision in the wake of a run of abuse scandals in state-run homes, Pindown in Staffordshire, the North Wales homes, Frank Beck in Leicestershire. Privatisation was, in part, a flight from the failures of the public model.

That does not absolve the sector, but it should temper any assumption that public ownership is inherently safer or better.

The empirical case has been put directly. The Competition and Markets Authority's 2022 study of the children's social care market found that for children's homes specifically, the costs of private and local-authority-operated provision were "broadly equivalent." The CMA declined to recommend a profit cap on a different basis: it judged that caps "would reduce incentives for providers to invest in services and cut capacity further" in a market with already-scarce placements. If the argument against private provision in children's homes were primarily about cost, the most thorough public investigation we have did not find that case.

More recent evidence pushes the picture further in the opposite direction to the public narrative. PSSRU's unit-cost work for 2021 puts the average weekly cost of an LA-run children's home at £4,865 against £4,153 for the independent sector, a 17 per cent premium. The Institute for Government's 2025 Performance Tracker puts the gap at over £750 a week.

Compliance does not run the way the narrative suggests either. Care Inspectorate Wales released enforcement data under FOI in September 2025 covering 1 June 2024 to 10 August 2025. Across that 14-month window CIW issued 38 Priority Action Notices to independent providers in Wales and 13 to LA-operated providers. Per home, the LA rate runs approximately twice the independent rate. The Operating Without Registration data is starker: 69 occurrences for LA-operated homes against 2 for the independent sector. CIW's own caveat is that these are counts of occurrences, not of children placed; some LA cases are likely fast-expanding services running ahead of paperwork. The 69-to-2 scale is still hard to dismiss.

None of this is adjusted for case complexity, which only widens the picture. DfE's 2020 analysis of needs profiles found that 90 per cent of homes accommodating children with mental health problems are privately run; LA and voluntary provision is over-represented in physical disability and short-break work, arguably the more predictable end of the sector. Average placement distance is 8 miles for LA placements and 43 miles for independent. The independent sector is running the difficult end of the sector, receiving more scrutiny per home, and still performing better on cost and compliance per placement.

The argument against private provision is therefore largely ideological. PE is not evil by default, and the relationship between ownership and child outcomes turns out to be more complicated than the front-page coverage suggests. Andrew Rome's April 2026 analysis of the top twenty independent providers found that several of the large registered operators explicitly refuse to offer unregistered placements because the reputational risk to their core registered business is too high. Profit motive is not the discriminating variable in the unregistered crisis.

Tucker's second distinction: not all unregistered provision is the same thing either.

The third category is where the harm concentrates. The policy debate mostly addresses the first. One further fact, which halts conversations when raised in sector forums: some local authorities contractually require providers to show a profit or surplus as evidence of financial stability. The public narrative about profiteering is aimed at something the same commissioning authorities have built into their own contracts.

It is worth being specific about how the sector operates, and how press coverage interacts with it.

A commissioner in the West Midlands recently described the following to a practitioner in contact with this publication. A new provider submitted a framework bid at £20,000 a week. The local authority called them to say the figure was well above the going rate. The provider's response: that was about the norm, a certain vocal journalist had been tweeting about it. New entrants to the sector are arriving off the back of these damning reports; the coverage encourages the very behaviour it seeks to vilify.

The mechanics of entry are low. One framework contract in the North West requires no physical building to tender; a Companies House registration and documentation completable in a day puts a provider on an approved list. Every provider on that list receives the same email when a child needs placing; the bid goes to the fastest or cheapest response. The number of registered children's homes in England doubled from 2,209 to 4,455 between 2016 and 2024, against a 9 per cent rise in the number of children in care over the same period. As buy-to-let landlords exit the rental sector, an advisory industry has emerged to walk them through the conversion into children's homes; one middleman markets it on Instagram as a "fully hands-off investment with guaranteed income and no ongoing headaches."

The mechanism that produces unregistered demand runs deeper than opportunistic operators. Research published by Public First and Commonweal Housing in April 2026 describes it as sequential sourcing. LA commissioning works through cheapest-to-most-expensive options in order: in-house fostering first, then independent fostering, then lower-cost residential, then specialist residential. By the time a child with high support needs reaches the part of the process where specialist homes are considered, those homes have been given a few days' notice and decline emergency referrals because of regulatory risk. "When it gets to the 27th day," one practitioner told the researchers, "the local authority has no option but to go out and look for unregistered provision." The last resort was always the only option. The Children's Commissioner estimated the annual cost of these placements at £440 million in 2024.

Underneath the sourcing pattern is a deeper failure of language and data. The sector has no agreed ontology for what good provision looks like, what specific quality components cost to deliver, or what outcomes a particular profile of child should be making progress toward. Specifications cannot ask for what cannot be named; contracts cannot pay for what cannot be measured; inspection cannot test for what nobody has defined. This is the children's-services equivalent of the market for lemons. What the system has built instead is a low-resolution risk scoring system resulting in a one or two word rating.

In April 2026, the Bureau of Investigative Journalism published the case of Alice, a 15-year-old placed by Caerphilly Council in Wales with MAP Adventures, an unregistered provider. The two members of staff caring for her, Liam Ramsay and Stephen Hurst, had seven criminal convictions between them, including four for violence. Both had signed documentation acknowledging Alice was at risk of child sexual exploitation. They were the people managing that risk.

Caerphilly Council placed Alice with MAP Adventures knowing it was an unregistered provider, and subsequently asked the company to continue holding her past the point at which the arrangement should have ended. She was attacked shortly afterwards. MAP Adventures went into liquidation. Caerphilly Council declined to give a public statement.

This is a chain of local authority decisions. Whether registered specialist provision was genuinely available to Caerphilly at the time, we don't know, and as this piece argues, by the crisis point it often isn't. But the decision to use an unregistered operator, and then to extend it, was the council's. No one in that decision chain has so far faced accountability. MAP Adventures was a private limited company, but profit was, in this particular case, beside the point. What the system failed to do was protect a 15-year-old girl in the care of two men with criminal records, whose presence in her life was authorised at every step by a local authority that knew what it was doing.

There is an accelerant in this system that almost never appears in press cycle analysis. It operates directly on the children most likely to end up in unregistered provision, and it involves the regulator as an active participant.

Other directors have described having to "risk-assess" Ofsted inspections themselves, that an inspection done badly causes some of the greatest harm in a child's life, destroying registered teams and managers with sweeping judgements drawn from snapshots that cannot capture the love, sweat, and tears that go into turning a child's life around. One director also asked me: "What happens to the kids after they are forced to move on?" He asks because he already knows the answer.

How the regulator got the job matters here. I've covered this in a separate piece; in short, Ofsted is an education inspectorate, given children's social care to inspect in the 2007 regulator consolidation, and politically engineered after Baby P into a buffer ministers can point to when accountability needs somewhere to stop. Successive governments have left the architecture in place. The Reset continues that pattern: Social Care Common Inspection Framework updates are promised for 2026 and 2027; no review of the institution itself is on the table.

The operational consequence has been documented. Public First found that in the mid-2010s, a request for an unregistered placement would cross a Director of Children's Services' desk roughly once every six months. A decade later, it is closer to at least once a week. The average duration of an unregistered placement is now 184 days. Thirteen per cent of the children in unregistered placements in September 2025 had been there for over a year. The emergency solution that was meant to last weeks is, for a significant minority of the most vulnerable children in England, where they live.

For me, this is the most important chart in Ofsted's "Good Decisions" report. The same report also documents that about a third of children's homes, 21 per cent "often" and 8 per cent "always," have to reject referrals for children with complex needs because of issues recruiting staff. The homes nominally best suited to those children cannot staff the placement; the children are pushed into unregistered settings the same regulator is now hiring a Senior Prosecution Lawyer to fine. The system creates the demand for illegal provision and then prosecutes the providers who pick up the children the system pushed out.

What is happening in the gap is this. Homes are not consciously lying. Safeguarding concerns, staffing ratios, risk to other children are real factors. But Ofsted fear sets the threshold, deciding when "enough" concern becomes a rejection.

The over-responsiveness to notifications has a specific institutional origin. In 2021, Ofsted closed a group of homes in Doncaster run by Hesley after sustained abuse of 106 children with learning disabilities. Ofsted had been notified of concerns on more than 100 occasions before the homes were finally closed. The regulatory response was to weight notifications heavily.

The homes most willing to take the hardest referrals are the ones that generate the most notifications. The mechanism that responded to Hesley is now the mechanism that pushes the hardest referrals out of the registered system and into unregistered provision. The system has created the very thing it claims to be most concerned about. Research in Practice, commissioned by Ofsted, found in July 2025 that the inspection framework cannot differentiate a provider doing hard therapeutic work with a child in active crisis from one warehousing the same child. Both generate notifications. Both face downgrade risk.

The children she is describing are overwhelmingly the ones the inspection framework has made it hardest to keep in registered settings.

In late 2024, a children's home director in the north of England had held a girl under twelve for over a year, the longest placement of her life after thirty prior placements. The director had embedded a head of therapeutic services and had leant in when nobody else would. A covering inspector visited, did not engage the therapeutic lead, instructed coercive management approaches, and downgraded the home to Inadequate, communicated by phone to the social worker rather than to the provider. The registered manager left the sector that day. The girl regressed. The placement broke down. She was moved to unregistered provision at over £22k a week. Ofsted re-evaluated the grade after it was raised, later reverted it, and sent a written apology letter.

Who won there? Not the child. Not the home that held her longest. Not the taxpayer paying the new weekly rate for a provision nobody inspects. The system that has a problem with unregistered provision had just created another instance of it.

Coverage of unregistered provision runs almost entirely on the Alice end of the range. It is not wrong to focus there, but it is incomplete in ways that matter for what would actually help.

Former Ofsted regulatory managers describe being personally involved in cases at the other end. A well-resourced registered home, impressive property, trained staff, full documentation. Inside, a boy kept in conditions one former manager described as virtually a cell: no deprivation of liberty order in place, staff scared of him. Ofsted suspended the registration immediately. No local authority home would accept him. He was placed in an unregistered setting with qualified nursing staff, a therapeutic support model, and a family-style environment. Within weeks he was having contact with his mother and wider family and making significant progress. A registered home was the danger. An unregistered home was where he recovered.

The assumption embedded in most coverage is that registered provision is the safe, less expensive option and unregistered provision the expensive and dangerous one. The Titheradge piece showed Staffordshire County Council paying £2.6 million last year for a registered, Ofsted-inspected placement requiring up to five staff, with the NHS sharing the cost. Few have called it a scandal in the way unregistered placements have been. The average registered placement now costs £6,100 a week. Whether the framing of the debate - registered equals safe and affordable, unregistered equals dangerous and exploitative - can survive the reality of what specialist provision actually costs at the extreme end of need, is the question.

Research commissioned in the South East of England is reported to have assessed several hundred children against CANS (Child and Adolescent Needs and Strengths) criteria. Practitioners familiar with the purported findings describe two cohorts: children in residential care whose needs would be better met in foster care, and children in foster care whose needs would be better met in residential. A significant number of looked-after children in England are in the wrong type of placement in both directions, and policy is currently planning the workforce on the assumption that none of them are. A national target of 10,000 new foster placements assumes the only problem is supply. (A longer piece on fostering is coming.)

The argument is not that unregistered provision is acceptable. It is that registration status and welfare outcome are not the same variable. Policy built on that conflation will keep producing Alice alongside cases where the registered sector was the danger and the unregistered setting was where a child recovered.

From April 2026, all new providers of children's residential services in Wales must be not-for-profit. It is the most significant supply-side reform in children's residential care in the UK in a generation. The intended logic is coherent: remove the profit incentive, remove the cowboy. The problem is that 87 per cent of Welsh children are currently placed with for-profit providers, and there is no funded alternative model to absorb them.

Alice was placed by a Welsh council with an unregistered provider that happened to be a for-profit limited company. The reform does not address the decision a Welsh local authority took to place a 15-year-old with an unregistered operator. The variable that put Alice at risk was not the registered for-profit sector the reform restructures.

Dr Kevin Gallagher runs registered homes in Wales and has been involved with the sector for over two decades. He named this risk before the mandate passed: the reform removes one variable while leaving the conditions that produced cases like Alice intact. Specialist therapeutic providers on thin margins exit first under regulatory uncertainty, before extractive operators do, because regulatory uncertainty is more expensive for a small specialist than for a group with diversified income and balance-sheet depth. Providers we speak to confidentially are already pulling out of Wales. The cost of the policy is falling on the children already in stable placements with the providers the policy is pushing out.

Researchers at Anglia Ruskin University, in a scoping study funded by the Eastern Region Special Operations Unit, identified the "borderless safeguarding" problem: a child placed from Cardiff into an unregistered setting in Bristol falls through every inter-agency monitoring system simultaneously — losing their local authority, their Ofsted jurisdiction, their IRO continuity, and their county lines visibility in a single placement decision.

The Wales reform, if it accelerates cross-border placements, does not solve the safeguarding gap. It widens it.

Ofsted advertised for its first permanent Senior Prosecution Lawyer in April 2026. Grade 6. £85,850. The regulator has had powers to prosecute unregistered providers since the 2021 regulations came into force. Zero successful prosecutions in five years.

The structural reason is less discussed than the number. The local authority that commissioned the illegal placement becomes the prosecution witness against the provider. The party that created the demand for illegal provision is also the party whose evidence is needed to close it down. That is not a failure of enforcement. It is a failure of design.

The new civil fining powers in the Children's Wellbeing and Schools Act require a separate commencement order from the Secretary of State, not expected before 2027. The law has passed; the power does not yet exist. Even before the Act received Royal Assent in April 2026, Cornwall Council had publicly tendered £6 million worth of placements with providers operating outside the registration framework. The council withdrew the advertisement after intervention from the Department for Education and Ofsted, and described it as a "soft market test, not a procurement exercise." Nobody was prosecuted.

The implementation plan does a great deal: Ofsted fining powers from Summer 2026, a financial oversight scheme for "difficult to replace" providers, new powers to refuse registrations on location and parent-undertaking grounds, expansion of Regional Care Cooperatives, statutory Family Group Decision Making, agency-worker regulation, and £560 million in capital for new open and secure homes. It contains the line that has had the most attention: a commitment that the government will "not flinch from capping the profits of private providers placing vulnerable children in care."

Four things the Reset does not do, all of which trace to the same underlying failure: the system cannot tell what a specific child needs, does not commission for need, and procures on price against a grade produced by a regulator that was never built for this job.

The Reset is not the wrong document. In many places it is the right document. But it is operating one layer above the place where the unregistered crisis is actually generated.

Public First and Commonweal Housing frame their April 2026 conclusion plainly: unregistered children's homes are "less a rogue outgrowth and more a symptom of how the system behaves under stress." They are a shadow system of care for the most vulnerable children in England, used not for routine cases but for those most at risk, when the regulated system cannot respond quickly enough.

Dr Mark Kerr called the sector this week a "Wild West" and the unregistered crisis "the culmination of 10 years of systemic failure to develop specialist provision." From several extensive conversations with professionals in the sector: every local authority has run the same policy for three decades - eradicate residential, prioritise kinship, then foster care, definitely not private, definitely not residential. Josh McAllister's care review repeated that policy with a bigger budget and more urgency, and specifically scoped out commissioning, the mechanism actually generating unregistered demand. The Reset is now the third iteration of that direction.

The Reset will be delivered. Most of the four reforms will land where the plan says they will. The 10,000 fostering places, the Kinship Zones, the new Regional Care Cooperatives, the financial oversight scheme, the Ofsted fining powers - these are real, they will move, the implementation plan is a coherent operational document.

What none of it reaches is the fact that the sector cannot, today, reliably tell what any specific child needs.

Without a shared ontology of need, commissioning runs against price and a grade, not need. Without commissioning against need, contracts cannot pay for the right thing. Without contracts paying for the right thing, inspection cannot test for it. Without inspection testing for it, a Dashboard reports against outcomes the sector has not yet agreed how to measure. Every layer of reform from here to 2029 sits on top of a missing definition.

I believe the instruments to fix this exist. The breakout, if there is one, is not a new regulator, a new Act, or a new fund. It is the agreed language for what a child needs, the discipline of commissioning against that language, and the data architecture that lets us see whether the children we are responsible for are moving in the right direction. Build the ontology first; commissioning, contracts, inspection, and the Dashboard all follow from it. Skip the ontology, and every reform from here on is reforming around a defect nobody has yet had the discipline to define.

And somewhere in a home that nobody inspects, a child whose longest stable placement ended with an inspector's phone call is still waiting for someone, anyone, in the chain of decisions that put her there to answer the only question worth asking: will this reform work?

Please share with a friend / colleague

Marcus (LinkedIn)