Every children's home in England gets a grade. Does the system care for the children inside them?

Ofsted inspects the home, on its own day, against its own framework, and most of the time it does that to satisfaction.

But the grade belongs to the building, not to the child who lives there. A child can be moved eight, ten, twelve times before they turn sixteen. Every home / fostering agency they pass through can be rated Good. Each decision is signed off on its own, one at a time.

Nobody is accountable for the line that runs through all of them

Whether the moves are meeting the child's needs, or just managing the next crisis. The home gets a rating, the placement itself gets risk assessed, but the trajectory does not get considered.

So a child can fall through a system where every single part is rated Good. Not because anyone gamed an inspection, but because we measure the home.

The child is what matters, and the two never meet in the data.

Now let’s briefly look at the other side of that placement decision.

A child needs to move by Monday. The home-finding team starts where cost is lowest: working from in-house foster carers, all the way through to the spot lists for private providers off framework. Those lists have grown so long that one authority I spoke to has had to split them into EBD1, EBD2, EBD3, EBD4…. just to be able to get the emails out. The same provider can sit on several of these list across different groupings.

So the referral goes out the way every authority sends it. As someone put it to me: "throw it at everybody and hope it sticks on something. It's a terrible model, but everybody does it." Nobody has a bed, or nobody has the right bed, or nobody can tell which, because the only information on file is the last inspection report. They make calls. Most go to voicemail. Two providers are full. One might have space in a fortnight.

They keep calling…

Meanwhile John Smith (made-up), who runs ABC Care in the Midlands, is getting around 700 referral emails a week. Maybe 45 to 60 a month are a genuine fit for the children his homes are registered and staffed to support. The other 640-odd are the wrong need, or a solo request for a child who does not need to be alone.

The home-finder cannot see that John has space. John cannot make himself visible in the noise (well he does but it sits in the vacancy email folder with 200 other emails).

The child waits.

The sector is not short of data. Ofsted holds a register of every home. Every authority holds child history. Every provider holds an inbox of referrals (700 a week). What nobody holds is the live picture: who with a vacancy, can take this child, keep them in school, and meet their needs, this week.

So yes I stole this quote from John… The data is everywhere. The intelligence to act on it does not exist.

Ofsted holds an extraordinary amount of this data. The registration details of every home in England. Inspection reports across thousands of them. Notifications on placements made and notices served. A survey of 8,880 children living in care. Research it commissioned itself, documenting most of the failures in this piece.

Some of it is public: the register, the inspection reports, the survey, the research. Much of it is not. The notifications in particular sit inside Ofsted and go no further (that i’m aware of).

But i’d argue publication was never the missing piece. The missing piece is that nobody (again im assuming, from many conversations), inside Ofsted or anywhere else, puts the three things that matter next to each other:

Joining up those things is the most obvious thing to me that has never been built. Ofsted inspects the home, but the system does not follow the child.

Ofsted's current design is not an oversight, I wrote on this previously. It is a structural decision that hardened after Baby P, when politicians needed a place for accountability to stop. They found it in a local authority: Haringey's director of children's services was publicly removed, sacked on the instruction of the Children's Secretary and announced live on television, a removal the Court of Appeal later ruled unlawful. Ofsted itself was widely judged to have escaped the blame. A tougher inspection regime followed. Every government since has treated the institution as politically untouchable (my personal take).

You can see the same pattern in the cases that reach the news. When a child dies, the inquest rarely finds one person who failed. It finds the gaps between them. In the case of one teenager who died in a children's home, the coroner's conclusion was a breakdown in communication between agencies, meetings the right people did not attend, and a system that leaned on the child to disclose what was being done to him.

No single part was accountable, because no single part could see the whole.

The result: Ofsted has been documenting the failure of children's social care information infrastructure for years. The regulator is simultaneously the best witness to the problem and the institution structurally unable to act on it.

That is it really (I’ll stop bashing Ofsted now)… the research is there, the data is there, what has always been missing is the infrastructure to connect them.

One provider with around 13 homes examined 100 randomly selected referrals in a single day. 27 asked for solo provision. Their referrals manager, a former registered manager, re-read all 100 and assessed a much smaller portion as genuinely needing solo placement. The rest had solo requested as the default.

All else equal, providers build what the typical referral asks for. From the data I can see bed counts per home have been shrinking year on year (I appreciate there are some larger legacy home structures). The sector is being blamed for a structure it built to specification.

Then there is the quality of the information itself. Under pressure to place, referrals first are - verbosed, let’s say - to get sign off, then get ‘softened’ ahead of circulation. I’m going to go out on a limb here and say the detail that would better route a child to the right home is exactly the detail that gets softened, because naming it plainly shrinks the knee jerk list of homes willing to say yes. I know this is a challenging topic, but a referral that says "fire-setting" could mean a child who lit a small garden bonfire with a friend years ago, or a child who set fire to their bedroom last month. Those are different children needing different homes, and burying the detail hides which.

Context matters.

Now let’s follow a HSB referral to the one provider, who could actually help. A specialist in supporting those with these needs knows precisely which children they are set up for. But they are reading 700 emails a week, scanning for the detail that let’s them say "this is one of ours," and that detail is the thing the referral is incentivised to obscure. So the best-matched home never opens the email, and the placement goes somewhere worse.

The referrer knows what the child needs and softens it, so the child is not turned away. The provider knows what it can really do and inflates it, so the referrals keep coming. Both sides bury the detail that would make the match better, and the placement gets decided by what is left in plain sight: distance, price, and who picks up the phone. Data rich, intelligence poor, on a single child (It is the market for lemons, flipped).

And it fails twice. The right routing is not just this week's email. It is the years before, when nobody told the market that this authority would keep needing homes for harmful sexual behaviour in this area, so no provider built for it. The signal is degraded at the moment of crisis, and it was never sent in the window before.

Ofsted's "Good decisions" report names the condition every referral decision depends on: a home's statement of purpose "accurately reflecting their current ability to care for children." 91% of local authorities surveyed said they often or always struggle to find suitable homes for children with the most complex needs.

In practice the providers can end up closer to marketing. With lack of clarity of what’s needed and where homes describe themselves as able to meet almost any need, because narrowing the description narrows the ask to them. Ask a provider what makes them therapeutic and, as one commissioner told me, the answer is often "we do PACE", a parenting approach she treats as the baseline every home should meet, not a specialism worth a premium.

The word "therapeutic" then justifies a therapeutic price?

Underneath it, nobody is checking the claim. A psychotherapist and a psychologist are different professionals with different training, and which one a home actually employs is unclear. In the medical system you could not advertise a clinical capability you could not staff. In children's social care from my understanding, you can, and the document meant to stop you is the one being used to do it.

Ofsted named the gap. Nothing followed.

A former LA commissioning manager I have been speaking with, who has spent 30 years on both sides of the fence, described what a fix would look like. Her framing: "Imagine a system like Rightmove. Providers put minimum information about their homes, type of location, type of needs they can meet. Local authorities do searches. In the background, it is anonymously collecting information about ages, needs, assessments. Over time you get the picture."

By now I have heard the Rightmove comparison from at least five people: policy authors, providers, commissioners, consultants…. everyone reaching for the same picture tells you the need is real and obvious.

But the picture is also the trap, because this is not Rightmove.

Rightmove works because the underlying data is simple and already structured: bedrooms, square footage, postcode, price, a few photos. Matching a child to a home is none of that. The information is unstructured and written in prose, it is patchy and needs enriching before anyone can trust it, and the match is not "does this fit the filter" but "is this the right placement for this particular child, with these needs, this week, near these people." On top of that sit frameworks, statutory duties, procurement rules and local politics no property portal ever had to touch.

Everyone can see why you would want it. Almost nobody builds the thing that actually holds at the speed, and under the pressure, it has to work.

She then listed every portal attempt she has watched fail. The DfE's own national attempt. Linkmaker, the live matching platform that already runs Scotland's national adoption register, and that Sir Martin Narey's 2016 review of residential care pointed to as a way of cutting the time wasted finding placements. It was never scaled nationally for children's homes. Local authority attempts covering only their own boundaries. The consistent finding across all of them: nobody brought providers in from the beginning, and nobody built it at national scale.

Ofsted commissioned research, carried out by Research in Practice at the National Children's Bureau, into how the sector conceptualises the children it serves. More than 400 participants. Published July 2025. The finding is in the title: "From trait to state: how Ofsted might consider conceptualising vulnerability for inspection and regulation."

The sector treats vulnerability as a trait. Something a child has. The paper's proposed framing: vulnerability is not a fixed trait but a changing state that everyone experiences.

This is not semantic. A trait is fixed. You design provision around managing it. A state is responsive to environment, support, relationships, and time. You design provision around changing it.

Commissioners commission against trait descriptions. Providers are implicitly paid to contain, not to change?

Sit with what that means for a single child. Nobody holds a full longitudinal view of them. Where they started, where they are now, where they are heading and why. What is working and what is not. What has worked, or failed, for other children who arrived looking like them. None of that is assembled anywhere. We have a system rich in single-day judgements about buildings and almost blind to the line a child travels through them. The grade tells you the home was in order on the day. It tells you nothing about whether the child is getting better.

The inspection framework still measures a home on a single day, without accounting for who is in it, without measuring whether the children in it are changing.

Look again at that survey of children in care. In children's homes, 94% said they feel safe. That deserves to land. But around 1 in 4 said they only sometimes or never get on with the others they live with, and close to a quarter do not always get to see the people who matter to them.

The grade and the lived experience are measuring different things.

Ofsted asked the question. Ofsted commissioned the answer. The inspection framework did not change.

No single individual can see the mismatch. Local authorities see their own demand (questionable!), one voice in 153. Providers see their own referral flow. Ofsted sees individual homes, registrations, inspections, notifications at points in time.

The result is in every dataset Ofsted publishes. The North West holds 26% of England's children's homes but only 18% of its looked-after children. Some London boroughs host far more children's home places than they will ever use for their own children, while others host almost none. 13 secure children's homes in England against 1,439 applications to deprive a child of their liberty in 2025. Generic solo EBD provision overbuilt in the north. Specialist provision chronically under-supplied everywhere it is needed.

Mid size operators read the commissioning tea leaves, build what referral volumes suggest is needed, generic EBD, often solo, open into a market that already has too much of the same, carrying vacancies. The ones that fill are those responding to actual complexity: CSE-experienced teenage girls in specific geographies, boys with harmful sexual behaviours in the Midlands. They fill because the provision matches the need.

The need was always there. The signal was not.

One provider told me how they stopped trusting the council guidance altogether. An authority had encouraged them to open a home for children with learning disabilities in the south, on the strength of what the authority believed its need looked like. The home opened. The referrals did not come. It ran under-occupied until they shut it. The reality the authority described was not the one that existed.

So the provider did something telling. Rather than ask the them again, they paid someone to work through several months of referral they had received, by hand, and rebuild the real demand from the bottom up: what this authority and its neighbours actually kept asking for, in what volumes, for which children. They built their own picture because the official one had cost them a home.

That is the state of the art for a provider trying to plan. Pay a person to read thousands of emails, because the system that should tell you is the one that got it wrong.

The data exists. The picture it would create does not.

Other systems have been stuck in the same place and found a way out. Not because what they do is anything like supporting a child, but because they faced the same information failure: the people who needed to find each other could not see each other.

I warn you in advance im going to talk about very abstract concept compared to children in care, so please be professional about it!!

Bond markets used to run on bilateral phone calls. The dealers held the information, the buyers did not, matching was slow and pricing was opaque. The fix was not more regulation. It was an terminal built by a company (Bloomberg) the information gap was expensive enough that people would pay to close it. The information layer arrived before the regulatory reform did.

The comparison ends at the information, and it has to. A bond can be moved to clear a market. A child cannot, and should never be spoken about as if they could. That is exactly why the burden has to fall on the information and not the child. You do not fix this by moving children faster, only by understanding each one well enough to get the placement as right as possible, the first time.

Healthcare ran on process compliance. Did you follow the checklist? A surgeon performing a hip replacement could document every step correctly and the patient could still be unable to walk six months later. Process compliance tells you whether rules were followed. It does not tell you whether the patient got better.

The fix was asking patients directly. How are you functioning now, compared to before? Then embedding those answers at commissioning level. You cannot commission for outcomes you cannot measure. Children's social care is still running on the checklist. The research that Ofsted commissioned named the frameworks that would capture state rather than trait. They have not been adopted at commissioning level.

What both fixes had in common was not that a private company did it. It was that someone built the missing information layer, and built it because the gap was costing people something they could not ignore. Who builds it matters less than whether it gets built at all, and in children's social care that is a live and uncomfortable question, because private involvement in the sector is already contested, and not always for bad reasons.

What we get instead of an information layer is policy lurch. The answer this year is to push hard toward fostering and lean against residential, much as an earlier year leaned the other way. These are large bets on the shape of provision, placed without the one thing that would tell you whether they are right: a clear, current picture of which children need what, and what is actually working for them. We keep changing the supply before we can see the demand.

The Regional Care Co-operatives are the most plausible forcing function for an information layer. On paper.

The case for yes is structural. A single RCC commissions at a scale where a matching layer is worth building, unlike any individual local authority. It has sufficiency strategy obligations to demonstrate under scrutiny from the Institute for Government and from ministers who have staked visible improvement on it. It also sits alongside the Children's Wellbeing and Schools Act 2026, which gives government new powers over provider finances, including financial oversight and a reserve power to cap provider profits. The Department for Education has told the National Audit Office it aims to improve cost transparency by 2028-29. Either way, providers will, for the first time, be publishing structured financial information into a regulatory system.

The case for no is the graveyard. Every previous attempt had structural reasons to work too. The DfE had national reach. Linkmaker had proven architecture. Local authority platforms had existing procurement relationships. Each failed for the same reason: nobody built it at the right scale, with providers inside the design, with a commercial incentive to make the data good.

The pattern that a clinician with years inside quality standards in therapeutic residential care described is worth sitting with: "We've gone through 'everyone needs to be attachment-informed'. Then 'trauma-informed'. These things are not new. They are repackaged conversations that arise at particular points in the cycle based on what is going on economically and politically." Each cycle generates new language and the same underlying failure reasserts itself, because the commissioning and measurement infrastructure never changed.

The RCC is the body that could break that cycle. I do not think it will, on its current approach, because it is reaching for the wrong architecture.

Look at what the sector actually buys when it tries to fix this. Authorities pay expensive consultants to rebuild a "needs matrix" every couple of years and still cannot tell you what their children need or where they are.

They run costing tools borrowed from adult social care that price a placement as if individualised care for a child were a unit of supply, which it is not. And the technology that gets commissioned is, more often than not, a procurement portal better suited to arranging pothole repairs than to matching a child to a home and critically following what happens next.

We are looking at a colourful picture through a black-and-white camera and wondering why we cannot make out the detail that matters.

The problem is not a missing report. It is the absence of an information layer underneath: children's needs, providers' real capabilities, and outcomes, structured well enough that they can finally be put next to each other.

That is the layer we are building with Bridge at elyndra. Not another portal, but the structured understanding every portal assumed was already there. Ontologies that turn a prose referral into something a machine and a stretched home-finder can both read. Enrichment that fills the gaps the official data leaves. Decision support that hands a commissioner a actionable shortlist instead of a spreadsheet with a column with initials in and another the weekly fee.

The conditions for getting this right are better than they have ever been. The track record of recognising what "this" actually is gives me reduced confidence the RCC will.

I would be glad to be proved wrong.

PS. Some salesy stuff… not the usual addition but i’m going to talk shop below…

If you commission, or you provide

I'll be at the commissioners' conference this week. If any of this lands, come and find me, I would love to discuss these challenges in person and always know there is more to this than what I currently know.

The question I'll be putting to commissioners is this: how many local authorities can show, at scale and on an ongoing basis, which children could be placed in-county, which need to leave, and why?

Without opening individual case notes. Without finding the answer on page seven of a referral. Without circulating that referral to 300 providers and hoping the right one notices.

Most cannot, and not because the people are not good. Because the information was never structured to make it possible. That is the part we have built. We take the referral data the system already produces, structure it the same way every time, and turn it into a picture of need: which children, with which needs, in which places, and how that is changing.

If you commission, we can show you your own sufficiency: what you could meet in-county, what you can’t and why, and where the gaps are before it’s becomes another. race against the statutory deadlines. 

If you provide care, we can help you understand the referrals you currently receive, how your placements process runs end to end and what you don’t see across the country, so you open and staff for what is actually needed (rather than for guesswork).

Either way the same structured data does the work.

The provider who talks to us moves toward more relevant referrals and less wasted time. The authority who talks to us gets a sufficiency picture it can act on. Both sides see more of the system than either can alone, and both carry less risk for it.

And this matters most of all:

It is how we get to sufficiency without the whole thing turning into a Punch and Judy show about ownership, profit, private equity and the fostering push.

Not that those questions do not matter, they do, but they come second to whether a child is actually getting what they need. And that order is the whole point. Put children's outcomes at the centre and measure them honestly, and the people who are not there for the children have nowhere left to hide.

You do not root the bad actors out by leading with ownership. You root them out by making it impossible to look good while failing a child. Get the matching right, and children start getting the specific support they need, without the drastic, system-wide moves that harm them along the way.

If that is a conversation worth having, reach out to me here [email protected], on LinkedIn or just say hello next week.

I make mistakes, and I know I don't know everything. Plenty of the people I talk to have spent more years in this sector than I've had birthdays…

What am I missing? I would genuinely like to know.

— Marcus (LinkedIn)