Safeguarding as Met or Not Met: Why It Is Never Blended with Anything Else
Why Ofsted's Inspection Toolkit judges safeguarding as a standalone threshold, what the decision science and safety research say about that design, and what it means for how a day nursery prepares, evidences and sustains its practice.
Written byAbraham
Published
Reading time17 min read
Under Ofsted's Early Years Inspection Toolkit, a day nursery receives graded evaluations across six areas: Inclusion; Curriculum and Teaching; Achievement; Behaviour, Attitudes and Establishing Routines; Welfare and Wellbeing; and Leadership and Governance. Safeguarding sits outside that grading scale altogether. It is judged separately as Met or Not Met. It is not averaged into the other areas, it cannot be offset by them, and it cannot be improved by strength elsewhere. For many practitioners this feels like an administrative detail. It is in fact one of the most deliberate design decisions in the framework, and understanding the reasoning behind it changes how a setting should prepare for inspection, how a leader should talk to staff about it, and how a safeguarding lead should structure evidence day to day.
This article explains the logic of keeping safeguarding apart from every other judgement, drawing on research from decision science, safety engineering, organisational psychology and child protection. It then sets out what that logic asks of a day nursery in practical terms. The central argument is simple: safeguarding is a threshold, not a score, and a setting should manage it as one.
The design logic: compensatory and non-compensatory judgement
Any system that combines several judgements into one outcome must decide whether strength in one area can make up for weakness in another. Decision researchers call a system where this is allowed compensatory, and one where it is not non-compensatory. Munda and Nardo (2009) make the point sharply in their analysis of composite indicators: when criteria are aggregated by weighted averaging, a poor result on one dimension can always be bought back by a strong result on another, and this is only defensible when the dimensions are genuinely substitutable. Where they are not substitutable, they argue, a non-compensatory approach is the only defensible setting. Averaging treats a safeguarding failure as though it were interchangeable with, say, a weaker mark on curriculum sequencing. It is not. No amount of excellent storytelling, outdoor provision or parent partnership compensates for a child being left at risk.
Psychological research on choice describes the same structure. Tversky's (1972) elimination-by-aspects model shows that people, when a particular attribute is non-negotiable, do not weigh it against others at all. They remove any option that fails it, however attractive that option is on every other dimension. A parent choosing a nursery behaves this way: a setting that is warm, well-resourced and affordable is still removed from the shortlist if the parent does not trust it to keep their child safe. Gigerenzer and Gaissmaier (2011) add that non-compensatory rules are not crude shortcuts. In uncertain environments, decisions based on a single decisive criterion often perform as well as, or better than, elaborate weighted models, because they are harder to distort and less vulnerable to compensating noise. A Met or Not Met judgement for safeguarding is the inspection equivalent of this: a decisive criterion that is deliberately kept out of the weighted mix.
The first lesson for practitioners is therefore conceptual. When you hear that safeguarding is judged separately, hear it as a statement about substitutability. Safeguarding is not one of the things the setting does well or badly. It is the condition under which everything else counts.
Why blending would fail: the Swiss cheese evidence
Safety research gives a second, independent reason. Reason's (2000) widely used account of organisational accidents describes a system as a series of defensive layers, each imperfect, like slices of Swiss cheese. Harm occurs when the holes in successive layers line up, creating a path through every defence. Two kinds of weakness create the holes: active failures, which are the unsafe acts of individuals at the front line, and latent conditions, which are the pre-existing weaknesses in systems, supervision, staffing and culture that make active failures likely and let them go uncorrected.
The relevance to grading is direct. If a framework averaged safeguarding with other areas, it would implicitly treat the defences as additive: a thick slice here compensating for a thin slice there. Reason's model says the opposite. A single hole that aligns with others is enough, and the thickness of the unrelated slices is irrelevant. Strong curriculum practice does not close a gap in recruitment checks. A warm key-person system does not close a gap in staff understanding of what to do with a disclosure. Blending would hide exactly the kind of local weakness the model identifies as dangerous.
In Reason's Swiss cheese model every layer of defence has holes, and harm happens when they line up. A thick slice elsewhere does not close a hole here. Photo: Wagner Vilas on Unsplash
Weick, Sutcliffe and Obstfeld (1999) describe how high-reliability organisations manage this in practice. They are characterised by preoccupation with failure, reluctance to simplify interpretations, sensitivity to operations, commitment to resilience and deference to expertise. In such organisations, small anomalies are treated as information about the system rather than as exceptions to be explained away. A day nursery does not need to be a nuclear plant to benefit from the same stance. A near-miss at pick-up time, a hesitant answer from a new member of staff about the allegations procedure, or a missed entry in a chronology are each a visible hole in a layer. A setting that treats such things as information is operating the way the framework hopes it will.
Guldenmund's (2000) review of safety culture research adds a necessary caution. What actually protects people is not the presence of procedures but the shared attitudes, assumptions and behaviours that determine whether procedures are used when it matters. A setting can hold a flawless safeguarding policy and still have a weak safeguarding culture. A threshold judgement, assessed through observation and conversation as well as documents, is a better fit for that reality than a score built from paperwork.
What serious case reviews tell us about single-point strength
The child protection literature supports the same conclusion from a different direction. Reviews of the deaths and serious injuries of children rarely identify a single cause. Laming's (2003) inquiry into the death of Victoria Climbié found failure spread across multiple agencies, with information held in several places and never connected. Reder and Duncan's (2003) analysis of communication in child protection networks found that the problem was seldom an absence of recorded information and much more often a failure to bring separate pieces of accurate information together. Sidebotham and colleagues' (2016) triennial analysis of serious case reviews for the Department for Education describes recurring patterns across very different cases, in which professionals had fragments of the picture and no structure for assembling them.
More recent work reinforces the point. Dickens and colleagues (2024) found that professionals frequently carried a concern for some time before it was formally raised, and that uncertainty about process, rather than lack of care, explained much of the delay. Taylor and colleagues (2024) describe the gradual normalisation of neglect, in which signs that would have alarmed a newcomer become unremarkable to those who see them every day. None of these studies describes a failing organisation that was otherwise excellent. They describe ordinary organisations in which defences were present but not joined up. That is exactly why safeguarding has to be assessed as a threshold: the question is not how much good practice exists around it, but whether, at the moment a child needs protecting, the system works.
The risks of a binary judgement, and how to manage them
A non-compensatory, binary judgement has real advantages, but it also carries risks that a thoughtful leader should name rather than ignore.
The first is indicator corruption. Campbell's (1979) classic observation, now usually called Campbell's Law, holds that the more weight a quantitative or categorical indicator carries in high-stakes decisions, the more it is subject to pressure that distorts the very process it is meant to monitor. A Met judgement is a high-stakes indicator. The risk is that a setting begins to manage the appearance of safeguarding, with immaculate files and rehearsed answers, rather than the substance. Perryman's (2006) work on performativity in inspected schools describes how staff under inspection pressure can come to perform compliance rather than live it, and Roberts-Holmes (2015) describes a parallel dynamic in early years, where practitioners learn to produce the data a judgement demands rather than trust their own observation. The antidote is not less rigour but a deliberate shift in what the setting counts as evidence: how staff behave when no one is checking, not how the file looks when someone is.
The second is defensive recording. Munro's (2011) review of the child protection system argued that a compliance-driven culture rewards the production of records over the exercise of professional judgement, and Parton's (2012) appraisal of that review warned that procedure can become a substitute for curiosity rather than a support for it. A binary threshold can feed this problem if staff conclude that the safest response is to write everything down and escalate nothing. A setting should be able to show not only that concerns were recorded but that someone read them, thought about them and acted.
The third is reliability. Bokhove, Jerrim and Sims (2023) found measurable differences between inspectors in the grades they award, and Ehren and Visscher (2006) show how strongly inspection outcomes can depend on how settings read and respond to the process. A binary judgement made on a short visit is not immune to such variation. What a setting can do is reduce the dependence on any one conversation. If safeguarding knowledge is genuinely distributed across the team, any inspector speaking to any practitioner will meet the same consistent understanding.
The fourth is the cliff edge. A pass-fail framing can make staff anxious in ways that reduce, rather than improve, safe practice. Edmondson's (1999) work on psychological safety shows that teams only surface difficult information when they believe they will not be punished for doing so. A leader who talks about Not Met as a catastrophe to be avoided at all costs may inadvertently teach staff to hide problems. A leader who talks about it as a minimum condition that is maintained by honest reporting of small failures teaches the opposite.
A worked illustration
Consider a hypothetical nursery, invented here for illustration only. Its curriculum is carefully sequenced, its key-person relationships are warm, its outdoor space is excellent, and parents speak highly of it. In a weighted model, these strengths would pull any overall judgement upward. Now suppose that, on a busy morning, a new member of agency staff is left alone with a group of children before the setting has seen proof of identity or confirmed the outcome of her suitability checks, because the usual manager was off sick and the cover arrangement was never written down. Nothing happens. No child is harmed. The hole in the layer is invisible in every other part of the setting's life.
Reason's (2000) model asks the uncomfortable question: if the right combination of other holes had been present that day, what would have stopped the harm? In a non-compensatory framework, the answer to whether the curriculum, the warmth or the garden could be counted in the setting's favour is simply no. Those things are real and valuable, and they are judged under their own headings. They are not currency that can be spent to cover a gap here. The nursery in this example would not be a bad nursery. It would be a nursery with a single unguarded point, and the framework is built to find single unguarded points.
The example also shows why the cause is usually latent rather than active. No one in the story acted maliciously or even carelessly. A cover plan was missing, a check was deferred under pressure, and each decision was reasonable in the moment. Weick and colleagues (1999) would describe the underlying problem as a failure to take small operational anomalies seriously, because nothing had gone wrong yet. Taylor and colleagues' (2024) account of normalised neglect has the same shape: it is the accumulation of small, explainable departures from the standard that eventually makes the unacceptable look ordinary.
Common misconceptions worth correcting
Several assumptions circulate in the sector that the research above does not support.
The first is that a strong overall profile protects a setting on safeguarding. It does not, by design. The areas are judged independently, and strength in one is not evidence about another. A setting that grades well on Leadership and Governance may still be Not Met on safeguarding if a specific arrangement fails.
The second is that safeguarding is the responsibility of one person. Reder and Duncan's (2003) analysis of failures in communication shows why this is dangerous: if information is only meaningful when it reaches the designated lead, every other member of staff is a potential point of failure. The lead coordinates; everyone holds the duty.
The third is that more paperwork means more safety. Munro (2011) and Parton (2012) both warn against this equation. A setting is not safer because its forms are longer, only because they are read, understood and used to prompt action.
A setting is not safer because its forms are longer, only because they are read, understood and used to prompt action. Photo: Sear Greyson on Unsplash
The fourth is that Met is a permanent state. It is a judgement on a particular day, based on evidence of arrangements and culture at that time. The only way to keep it true is to keep doing the things that make it true, which is why the research on safety culture emphasises continuous attention rather than one-off preparation.
What this means in practice for a day nursery
The research above converts into a small number of concrete habits.
Keep a standalone safeguarding evidence file that does not depend on any other area. It should show the setting's arrangements as a whole: recruitment and vetting, induction and training, the designated safeguarding lead and deputy, the chronology and referral process, supervision, allegations handling, online safety, visitor control and the setting's approach to attendance and missing children. Do not rely on the strength of the curriculum or welfare files to carry it. The framework does not allow them to.
Test the defences, not just the documents. Reason's (2000) model implies that the useful question is where the holes might line up, so run short scenario exercises in staff meetings: a child makes a disclosure at the end of the day; a parent arrives who is not on the collection list; a colleague's behaviour in the baby room causes unease. Ask each person what they would do, to whom they would speak and what they would write. The point is not to catch anyone out but to find the layer that has a hole.
Short scenario exercises in staff meetings find the layer with a hole: what would you do, who would you speak to, and what would you write? Photo: LinkedIn Sales Solutions on Unsplash
Treat near-misses as information. In the language of Weick and colleagues (1999), cultivate preoccupation with failure and reluctance to simplify. When something goes slightly wrong, record what happened, ask what in the system allowed it, and record what changed. A setting that can show a short log of small issues identified and fixed is demonstrating the behaviour the framework is looking for. A setting with nothing in such a log is not necessarily safe; it may simply not be looking.
Connect the information the setting already holds. Reder and Duncan's (2003) finding that harm so often follows a failure to connect accurate information suggests a simple discipline: at a regular interval, the safeguarding lead reads attendance, accidents, concerns, key-person observations and any early help or social care involvement together for each child of concern, rather than as separate records. If your register, accident book and concerns log are never read side by side, they are separate slices of cheese with no one checking whether the holes line up.
Make speaking up routine. Morrison and Milliken (2000) describe how organisations drift into silence when staff come to believe that raising a concern is pointless or risky. Mannion and Davies (2015) describe the same dynamic in healthcare. A setting that wants its safeguarding to be Met in practice needs a culture in which a room assistant can tell the deputy about a concern without hesitation, in which the answer is always a thank-you, and in which the person who raised it is told what happened. That is what makes the written procedure real.
Prepare people, not scripts. Because inspectors speak to practitioners as well as reading files, the best preparation is not rehearsing answers but ensuring every member of staff, including bank and agency staff and students, can explain in their own words what they would do if a child told them something worrying, and can name who the safeguarding lead is. Understanding that survives a change of phrasing is what a threshold judgement tests.
Finally, talk to staff about the judgement honestly. Explain that safeguarding is judged separately because it is the condition on which everything else depends, and that the setting's aim is not to avoid Not Met but to stay far enough above the line that the question never arises. Present it as a shared professional responsibility rather than an external threat.
Conclusion
The separate Met or Not Met judgement is not a bureaucratic quirk. It reflects a body of evidence about how non-substitutable criteria should be judged, how complex systems fail, and how children come to harm in organisations that thought they were doing well. Averaging safeguarding with other areas would reward the wrong thing and conceal the right one. For a day nursery, the practical consequence is to stop thinking of safeguarding as one more area to evidence well, and to start thinking of it as the foundation that the other five areas stand on. A setting that manages it that way, through honest reporting, connected information, tested defences and a culture where small concerns are welcomed, will not only meet the threshold on the day of an inspection but will be a safer place for children on every other day.
References
Bokhove, C., Jerrim, J., & Sims, S. (2023). Are some school inspectors more lenient than others? School Effectiveness and School Improvement, 34(4), 419-441.
Campbell, D. T. (1979). Assessing the impact of planned social change. Evaluation and Program Planning, 2(1), 67-90.
Dickens, J., Cook, L., Cossar, J., Okpokiri, C., Taylor, J., & Garstang, J. (2024). The multiple and competing functions of local reviews of serious child abuse cases in England. Critical Social Policy, 44(3), 447-467.
Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350-383.
Ehren, M. C. M., & Visscher, A. J. (2006). Towards a theory on the impact of school inspections. British Journal of Educational Studies, 54(1), 51-72.
Gigerenzer, G., & Gaissmaier, W. (2011). Heuristic decision making. Annual Review of Psychology, 62, 451-482.
Guldenmund, F. W. (2000). The nature of safety culture: A review of theory and research. Safety Science, 34(1-3), 215-257.
Laming, Lord H. (2003). The Victoria Climbié Inquiry: Report of an inquiry by Lord Laming. The Stationery Office.
Mannion, R., & Davies, H. T. O. (2015). Cultures of silence and cultures of voice: The role of whistleblowing in healthcare organisations. International Journal of Health Policy and Management, 4(8), 503-505.
Morrison, E. W., & Milliken, F. J. (2000). Organizational silence: A barrier to change and development in a pluralistic world. Academy of Management Review, 25(4), 706-725.
Munda, G., & Nardo, M. (2009). Noncompensatory/nonlinear composite indicators for ranking countries: A defensible setting. Applied Economics, 41(12), 1513-1523.
Munro, E. (2011). The Munro review of child protection: Final report – a child-centred system. Department for Education.
Parton, N. (2012). The Munro Review of Child Protection: An appraisal. Children & Society, 26(2), 150-162.
Perryman, J. (2006). Panoptic performativity and school inspection regimes: Disciplinary mechanisms and life under special measures. Journal of Education Policy, 21(2), 147-161.
Reason, J. (2000). Human error: Models and management. BMJ, 320(7237), 768-770.
Reder, P., & Duncan, S. (2003). Understanding communication in child protection networks. Child Abuse Review, 12(2), 82-100.
Roberts-Holmes, G. (2015). The "datafication" of early years pedagogy: "If the teaching is good, the data should be good." Journal of Education Policy, 30(3), 302-315.
Sidebotham, P., Brandon, M., Bailey, S., Belderson, P., Dodsworth, J., Garstang, J., Harrison, E., Retzer, A., & Sorensen, P. (2016). Pathways to harm, pathways to protection: A triennial analysis of serious case reviews 2011 to 2014. Department for Education.
Taylor, J., Cook, L., Cossar, J., Dickens, J., Garstang, J., & Okpokiri, C. (2024). Tackling the "normalisation of neglect": Messages from child protection reviews in England. Child Abuse Review.
Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79(4), 281-299.
Weick, K. E., Sutcliffe, K. M., & Obstfeld, D. (1999). Organizing for high reliability: Processes of collective mindfulness. Research in Organizational Behavior, 21, 81-123.
Operations Director, Awake Early Years Consultancy
Former Ofsted Early Years Inspector, doctoral researcher and early childhood education and care consultant with over 16 years' experience across practice, inspection and leadership.
