The working alliance is not a therapy-specific idea imported loosely into safeguarding. It is an established construct — a bond between practitioner and client, agreement on the goals being worked toward, and agreement on the tasks required to reach them (Horvath and Greenberg, 1989) — that has been directly measured and validated within child protection services themselves, including in UK samples (Killian et al., 2017). That foundation, however, cannot rest on trust in the therapeutic sense, because too many families arrive at the door having already had trust broken by people who mattered to them, often as an ordinary feature of difficult lives rather than an extraordinary one. Asking a family to trust a service risks reactivating that history. What can be built instead, deliberately and observably, is confidence — a reasonable, forward-looking expectation that this particular service, functioning as it does, is likely to be worth engaging with. Confidence, unlike trust, does not require the family to relitigate who has hurt them before. It only requires the service to behave, consistently, in ways that earn it.
The working alliance should be distinguished from the narrower, more familiar term it is often confused with: the therapeutic alliance. Bordin's original formulation (1979), from which Horvath and Greenberg's measure was later derived, was already deliberately broad — bond, goal, and task apply to any helping relationship, not only psychotherapy. The therapeutic alliance, by contrast, is usually discussed as the relationship within an actual therapeutic intervention, and its character shifts with the modality in question: a cognitive behavioural alliance tends to look collaborative and task-focused, closer to coaching; a psychodynamic alliance treats the relationship itself, including the transference within it, as a primary object of the work; a person-centred alliance holds the relational conditions themselves (Rogers' congruence, unconditional positive regard, and empathic understanding) as sufficient, in principle, to enable adaptation. Across all three, the alliance is frequently where therapeutic movement actually occurs — where symptoms ease or a shift from dysfunction toward function becomes possible — rather than being merely the backdrop against which technique operates. It would be a mistake, though, to treat this as available only to staff holding the job title of therapist. Social workers and family support workers are doing directly comparable work: engaging a family face to face, or managing the network of services around them, with the same underlying aim of preventing harm and moving a family from unhealthy toward healthier functioning, often with a clear focus on reducing specific difficulties in a child's or parent's presentation. That work carries a therapeutic alliance of its own, whether or not "therapist" appears on the practitioner's contract. The distinction that matters is not job title but modality and setting: a social worker's therapeutic alliance will look different from a psychodynamic therapist's, in the same way a CBT alliance looks different from a person-centred one, but there is no principled reason to treat it as a lesser or absent version of the same thing. Where a family also receives input from a separately titled therapist — as some do in the fictional Midkwell case material accompanying this suite — that practitioner holds a further therapeutic alliance of their own, related to but distinct from the one the social worker or family support worker is already holding, and neither should be assumed to substitute for the other.
The overlap is not only conceptual. When a social worker or family support worker sits with a family to build a genogram or an ecomap, uses role play or a dolls' house to help a child show what words cannot, or holds an in-depth conversation about a difficult past or recent experience as part of an assessment or a plan, they are frequently using the same tools a therapist would use, for closely related purposes. From the family's side of that conversation, there is often no meaningful difference in how it is experienced: a genogram exercise conducted by a social worker can surface exactly the same family history, the same buried conflict, the same grief, as one conducted by a family therapist. A child expressing something through a dolls' house with a family support worker is doing the same act of symbolic disclosure a play therapist would be trained to interpret. These sessions can be every bit as searching, and every bit as invasive, as anything that happens in a clinical therapy room — the fact that they sit inside a statutory assessment or a support plan rather than a course of therapy does not make them lighter for the person living through them. A practitioner without the job title of therapist is not exempt from the emotional weight this work carries for a family, and should not assume that the absence of a therapeutic label makes the encounter itself any less significant.
Confidence of this kind cannot be produced by policy statement, and it resists being faked for long by either party. A family only extends it where the service's conduct gives genuine grounds to; a team only earns it where its underlying practice — its resourcing, its cultural competence, its consistency, its capacity to act on what it observes — is sound. Confidence, and the working alliance it makes possible, is treated here as a dependent variable: something produced by identifiable features of good practice, worth naming individually, rather than a fortunate accident of an easy family or a personable worker.
One recurring, deliberately concrete image keeps the argument anchored: the door. Not a metaphor for anything vague, but the literal point of entry — a home visit, a school gate, a hospital ward, a first meeting in an unfamiliar room — at which a professional physically crosses into a family's own space for the first time, or the first time in a while. The door matters because of what is true on both sides of it at that exact moment, and at very few others: confidence, for worker and family alike, is at its lowest possible ebb. The worker does not yet know what is inside. The family does not yet know whether this visit will go the way the last difficult one did, or the way a good one might. Every description of good practice that follows is, in effect, a description of what happens to that shared uncertainty over time and repeated contact — whether it resolves into confidence earned on both sides, or hardens into something neither party can undo.
Meeting this subject through failure
Social justice content in social work education is often taught through case studies of how things go wrong. This is not a criticism of the method — the cases are real, the lessons hard-won, and no serious safeguarding education could responsibly skip them. But a curriculum built almost entirely from post-mortems carries a quiet risk of its own: it can teach someone to recognise dysfunction in fine detail while leaving them with only a vague, unexamined sense of what its opposite actually looks like. It is one thing to be told, repeatedly, what failure resembles. It is another to be shown, with the same precision, what success is made of.
The structural problems are named only as long as necessary to establish why building confidence, rather than assuming or demanding trust, matters at all. Attention then turns to what effective, socially just practice concretely achieves for a child and for a community, and, more importantly, what it is about that practice that produces the achievement — a mechanism anyone working in this field, whether newly qualified or long-serving, can carry into practice and use, not a longer list of cautionary tales to add to the ones they already have.
The shape of the problem, briefly
The starting fact is a counter-intuitive one. Research led by Paul Bywaters and colleagues at Coventry University, running since the mid-2010s, has repeatedly found that a family's chances of being drawn into the child protection system depend not only on how deprived they are, but on how deprived their local authority is as a whole, and how unequal that authority is internally (Bywaters et al., 2015; Bywaters et al., 2018). For any given level of neighbourhood deprivation, a child in a less-deprived, more unequal local authority is more likely to receive a child welfare intervention than an equally poor child in a more deprived, more equal one — a relationship strong enough, on some analyses, to run several times over between the least and most equal authorities. This has become known as the inverse intervention law, and it means, in plain terms, that two equally poor families can face very different odds of statutory involvement depending on nothing more than which side of a local authority boundary they happen to live on — an inequality no amount of individual alliance-building between one worker and one family can fix. That distinction matters for what follows, at two levels that need to stay separate. Structural justice — resourcing, policy, how a system is designed — sets the conditions under which a working alliance becomes possible or impossible at scale. Confidence and alliance, the practice-level focus of what follows, are what a team then builds within whatever those conditions allow, case by case. Good practice at the second level cannot substitute for justice at the first; it can only do as much good as the structure around it permits.
Poverty of this kind has been a consistent presence in the UK's serious case review literature for two decades, named carefully and without being treated as a direct cause of harm — the government's own triennial analyses of serious case reviews found poverty featured in roughly half of all cases reviewed between 2011 and 2019, usually as one pressure compounding others rather than an explanation on its own (Featherstone et al., 2019). Race has had a very different history in the same literature, examined in full in Race and the Licensing of Discourse — from Lord Laming's 2003 inquiry into Victoria Climbié's death, through Bernard and Harris's 2019 confirmation that race remained "limited attention, or virtually absent" across the reviews they examined, to the Child Safeguarding Practice Review Panel's own 2025 report, titled, without much room for misreading, "It's Silent."
Two further cases sit either side of this thread and complicate it usefully. Peter Connelly died in Haringey in 2007, in the same borough where Victoria Climbié had died five years earlier, after being seen by agencies sixty times in eight months. His was not a case of low visibility — his lead social worker's caseload had doubled in six months to fifty per cent above the recommended level, in a borough that was, at the time, the fourth most deprived in London (Haringey Local Safeguarding Children Board, 2008, 2009). The case prompted more than a local review: Lord Laming, who had chaired the original Climbié inquiry five years earlier, was brought back by the government to assess how far his own recommendations had actually taken root nationally (Laming, 2009) — a rare instance of the same reviewer assessing the failure of his own earlier reforms. Peter's case is a hard corrective to any assumption that contact alone builds a working alliance: repeated contact without the capacity to act on what it reveals earns nothing, however frequent it is. Rikki Neave, murdered in Peterborough in 1994 by another child and only convicted in 2022 after new evidence emerged, offers a different complication again. His mother was in frequent, documented contact with social services and was herself later convicted of neglect — this was not an unseen family. No formal case review of the kind now standard appears to have followed his death; instead, the Department of Health's Social Services Inspectorate was commissioned to inspect Cambridgeshire's child protection services, finding serious deficiencies and issuing forty recommendations (Social Services Inspectorate, 1997). Public commentary at the time leaned heavily on a moralising, class-coded reading of Rikki's mother as a chaotic single mother, an "underclass" framing common to the period, which sat alongside, rather than instead of, the actual practice failures the inspection identified.
None of this will be new to anyone who has read a serious case review. What is less often set out is what the alternative looks like in practice.
What good practice achieves for the child
Each of the outcomes below can be tested against the same question: does this give the family genuine grounds for confidence in the service, rather than merely an absence of complaint?
Contact that is not gated by contactability. A home visit during standard office hours remains, by a wide margin, the taken-for-granted way child protection contact happens (Ferguson, 2016). The neighbourhoods most often seeing child protection intervention tend to be those where someone is reliably home during the working day and answers the door — a pattern distinct from, and easy to mistake for, the actual level of risk present. Good practice builds in more than one route to visibility — school, health services, community contact, informal networks — so that a family's safety never rests entirely on whether an assessor happens to catch someone at home, and so that confidence in the service isn't quietly reserved for families whose working patterns happen to suit it.
Being seen without being presumed guilty. Effective practice keeps "this family needs support" and "this family is suspected" as genuinely separate questions, so that contact carries help at least as often as scrutiny. Where that separation breaks down, entire communities can learn to expect scrutiny before any assessment has taken place at all — precisely the dynamic examined in full in Race and the Licensing of Discourse, where the 2025 Panel's report on race and racism describes racial bias shaping decisions silently and unexamined.
Accurate reading of the whole child. Good practice avoids both of the failure modes the evidence above illustrates: the cultural relativism that may have let clearly alarming signs in Victoria Climbié's case go unchallenged, and the racial bias the 2025 Panel's report found operating in the opposite direction elsewhere. The Social GRACES framework — Burnham and Roper-Hall's tool, now extended well beyond its original seven factors to a fifteen-factor list still in active use in UK safeguarding training (Burnham, 2012) — is intended to help a practitioner hold several dimensions of a family's identity in view at once, moving between what is visible or invisible, voiced or unvoiced, depending on the moment. Used as Burnham intended, as what he called a "collide-scope" rather than a checklist, it is one of the better tools available for keeping both failure modes in view simultaneously.
Contact matched with the capacity to act. Peter Connelly's case states this lesson at its most stark: sixty contacts built no meaningful alliance and protected no one, because the worker holding his case did not have the caseload, time, or organisational support to act on what those contacts showed. Good practice is not simply "notice more" — it is noticing paired with the resourced capacity to respond to what is noticed, and a family has little reason for confidence in a service that sees without acting.
Continuity of relationship. A child who is not required to re-explain themselves to a new worker every few months, and who is not institutionally defined by whichever incident first brought their family to attention, experiences a materially different kind of contact. A 2026 scoping review of child protection worker turnover found that relational discontinuity directly weakens the working alliance and undermines the trust-building process between families and professionals (British Journal of Social Work, 2026) — continuity is not a courtesy, on this evidence, but a significant factor in whether the alliance forms at all.
What good practice achieves for the community
The same test applies at community scale: does this give the community as a whole genuine grounds for confidence in the service, rather than reasons to brace for or evade it?
Investment that follows need rather than tracking away from it. The most direct positive answer to the inverse intervention law is resourcing that actually matches deprivation, rather than one that inversely correlates with it. Where investment follows need, the postcode a family lives in stops determining their odds of statutory involvement.
The community stops functioning as a risk proxy. A postcode's reputation, or a group's historical over-representation in past reviews, should do no predictive work on an individual family that has not itself been assessed. The Panel's own 2025 figures on over-representation are worth reading, in this light, not as a stable feature of how the system works but as evidence that something upstream still needs fixing.
Relationship precedes crisis. Contact that is embedded and ongoing, rather than triggered only by crisis, is what actually allows confidence to accumulate in a community's lived experience, rather than remaining a reassurance professionals offer about their own intentions. The same 2026 turnover review found that shared ethnicity, kinship care arrangements, and social support acted as protective factors that strengthened the working alliance — direct evidence that structural, community-level conditions and the interpersonal alliance are not two separate stories, but the same one at different scales.
Services designed around how people actually live. This is where an observation about contactability becomes something genuinely solution-focused rather than merely descriptive. If households where every adult works fixed daytime hours are structurally harder to find at home for an unannounced or office-hours visit, the answer is not to wait for families to conform to a nine-to-five assessment model but to design contact patterns around how they actually live. Without direct professional experience of Gypsy and Traveller communities to draw on here, the following is offered only as a hypothesis worth testing rather than an established finding: some of these communities' more flexible, self-directed patterns of income-generating work may correlate with greater at-home availability during the working day, which — alongside other, better-established drivers of disproportionality, including service design that fails to accommodate itinerant or non-standard living patterns — could be worth examining as one contributor to the over-representation of Gypsy and Traveller children in safeguarding contact and in Child Safeguarding Practice Reviews. The point is not that any community should be more or less contactable; it is that a service designed around the rhythms of the people it serves, rather than around a single assumed default household, is a service with a genuine chance of earning confidence rather than merely being endured.
Stigma has somewhere to end. Collective judgement of "that estate" or "that community" — the same judgement that shaped public reaction to Rikki Neave's mother in 1994, and that the 2025 Panel found still shaping decisions along racial lines three decades later — does not survive a system that assesses the individual family in front of it rather than the statistic it happens to resemble.
What makes this practice good
A list of good outcomes is only useful in practice if it comes with some account of what produces them — otherwise it reads as a set of virtues to admire rather than a practice to build. A 2019 study observing 127 families found that relationship-building skills directly predicted parent-reported engagement on the Working Alliance Inventory, while "good authority" and the evocation of intrinsic motivation showed stronger relationships with outcomes still (British Journal of Social Work, 2019) — confidence, in other words, is not simply a matter of being liked, but of specific, teachable practice behaviours. Each of the following can be checked directly against that same test: does it give a family genuine, earned grounds for confidence in what this service does?
Redundancy of contact. Safety should never depend on a single channel succeeding. Multiple routes to visibility mean one missed connection does not become the whole story.
Separation of visibility from suspicion, built into the design of a service, not left to individual judgement in the moment. A worker's good intentions cannot substitute for a system that has already decided, structurally, that contact and scrutiny are not the same thing.
Caseload capacity treated as a safeguarding variable, not a budget line sitting apart from safeguarding outcomes. Peter Connelly's case makes the alternative unmistakable: a caseload that has doubled in six months is not a resourcing footnote to a safeguarding failure — in a very real sense, it is the failure.
Reflexivity tools used as intended. The Social GRACES framework, revisited across the life of a case as Burnham's "collide-scope" concept intends, does more work than the same framework completed once as an exercise and filed. A more theoretically rigorous account of how these factors compound — the intersectionality literature the GRACES framework has itself been critiqued against (Nolte, 2017, discussed in a 2021 review of the framework's strengths and limits, full citation still pending — see References) — is worth attention precisely because GRACES was never designed to capture everything alone.
Structural humility. Designing services around how communities actually live — their working patterns, family structures, and degree of settlement or transience — rather than requiring communities to fit an unstated default, is not a compromise on rigour. It is what rigour looks like once contactability itself is recognised as unevenly distributed.
Confidence built even where contact is mandated, not chosen. Child protection work is rarely voluntary in the way therapy is, and it might be assumed that a genuine working alliance is impossible under legal compulsion. The evidence does not bear this out: qualitative research within a family residential treatment programme found positive working alliances were achievable even within a mandated, coercive context, according to the parents themselves (Full article, Taylor & Francis, 2024) — a finding directly relevant to anyone entering statutory practice, where almost every relationship starts under some degree of compulsion.
The nexus, and the door
Social work education, in the words of one widely used curriculum, describes its own remit as introducing newcomers to "the nexus between power and knowledge, gender relations, class, families, communities, organisations, race and ethnicity and other societal injustices." Read alongside a stack of serious case reviews, that language can sound like a warning: here is everything that can go wrong, and here is the vocabulary for describing it. Read alongside what has been argued here, it can be read differently — as a description of exactly what a functioning system gets right, whenever it does.
Nobody needs a curriculum to learn what went wrong in the cases already documented. The cases make that plain enough on their own. What a curriculum like this can teach, read this way, is what going right is actually made of — resourcing that follows need, contact that carries help rather than only scrutiny, reflexivity that holds a whole family in view, services shaped around how people actually live rather than how a policy assumes they live. None of it is abstract once it is put this way, and none of it depends on individual virtue so much as on a set of structural choices any team can make.
A family that has learned, contact by contact, that this service gives it grounds for confidence, and a family that has learned the opposite — for reasons that may have had very little to do with them and a great deal to do with the system built around them — are not different in kind. Both started at the same door, in the same state of shared uncertainty. Neither family is being asked to trust again what has failed them before; both are simply asking, reasonably, whether this particular service has earned anything yet. The distance between a family that can answer yes and one that cannot is not a mystery, and it is not fixed. Some of that distance is structural, and no team can close it alone — the same distinction drawn at the outset, between what justice at the structural level makes possible and what practice then builds within it, still holds at the close. But a meaningful part of it comes down to identifiable, teachable, practice-level choices that either build a working alliance worth having or fail to, one contact at a time — starting, always, at that same door.
References
Bernard, C. and Harris, P. (2019) 'Serious case reviews: The lived experience of Black children', Child & Family Social Work, 24(2), pp. 256–263.
Bordin, E.S. (1979) 'The generalizability of the psychoanalytic concept of the working alliance', Psychotherapy: Theory, Research & Practice, 16(3), pp. 252–260.
British Journal of Social Work (2019) study on worker skills and outcomes in child and family social work, 127 families, Working Alliance Inventory as primary engagement measure — full citation (authors, volume, issue, pages) to be confirmed; located via Oxford Academic, BJSW, published online 1 December 2019.
British Journal of Social Work (2026) scoping review, 'Effects of child protection social worker turnover on children and families' — full citation (authors, volume, issue, pages) to be confirmed; located via Oxford Academic, BJSW, published 1 May 2026.
Burnham, J. (2012) 'Developments in Social GRRRAAACCEEESSS: Visible-invisible and voiced-unvoiced', in Krause, I.B. (ed.) Culture and Reflexivity in Systemic Psychotherapy: Mutual Perspectives. London: Karnac, pp. 139–160.
Bywaters, P., Brady, G., Sparks, T., Bos, E., Bunting, L., Daniel, B., Featherstone, B., Morris, K. and Scourfield, J. (2015) 'Exploring inequities in child welfare and child protection services: Explaining the "inverse intervention law"', Children and Youth Services Review, 57, pp. 98–105.
Bywaters, P., Brady, G., Bunting, L., Daniel, B., Featherstone, B., Jones, C., Morris, K., Scourfield, J., Sparks, T. and Webb, C. (2018) 'Inequalities in English child protection practice under austerity: A universal challenge?', Child & Family Social Work, 23, pp. 53–61.
Child Safeguarding Practice Review Panel (2025) "It's Silent": Race, Racism and Safeguarding Children. London: Child Safeguarding Practice Review Panel.
Featherstone, B., Morris, K., Daniel, B., Bywaters, P., Brady, G., Bunting, L., Mason, W. and Mirza, N. (2019) 'Poverty, inequality, child abuse and neglect: Changing the conversation across the UK in child protection?', Children and Youth Services Review, 97, pp. 127–133.
Ferguson, H. (2016) 'Making home visits: Creativity and the embodied practices of home visiting in social work and child protection', Qualitative Social Work, 15(2), pp. 153–170.
Full article, Taylor & Francis (2024) 'Establishing a positive working alliance during formal parenting assessments in a family residential treatment (FRT) program: Parents' perspectives on what is helpful' — full author names and journal to be confirmed before use.
Haringey Local Safeguarding Children Board (2008) Serious Case Review 'Child A': First Overview Report, November 2008. Published (redacted) by the Department for Education, 26 October 2010.
Haringey Local Safeguarding Children Board (2009) Serious Case Review 'Child A': Second Overview Report, March 2009. Published (redacted) by the Department for Education, 26 October 2010.
Horvath, A.O. and Greenberg, L.S. (1989) 'Development and validation of the Working Alliance Inventory', Journal of Counseling Psychology, 36(2), pp. 223–233.
Killian, M. et al. (2017) study validating the Working Alliance Inventory–Short within UK child protection services (London-based experimental study) — full citation to be confirmed before use.
Laming, Lord (2003) The Victoria Climbié Inquiry: Report of an Inquiry by Lord Laming, Cm 5730. London: The Stationery Office.
Laming, Lord (2009) The Protection of Children in England: A Progress Report, HC 330. London: The Stationery Office.
Nolte, L. (2017) as cited in a 2021 review of the Social GRACES framework's strengths and limitations, American Journal of Family Therapy, 48(6) — this is the single outstanding citation gap in the essay: the 2021 review's own author names are still needed.
Rogers, C.R. (1957) 'The necessary and sufficient conditions of therapeutic personality change', Journal of Consulting Psychology, 21(2), pp. 95–103.
Social Services Inspectorate (1997) Inspection of Child Protection Services in Cambridgeshire. London: Department of Health. [Follow-up inspection 1998 assessed implementation of the 1997 report's forty recommendations.]
Validity of the Working Alliance Inventory within child protection services (Cardiff/ORCA study) — 130 families, social workers on 274 cases, observers on 165 home visits, randomised controlled trial — full author names and journal to be confirmed before use.
Topics: #WorkingAlliance #Confidence #ChildProtection #Safeguarding #InverseInterventionLaw #SocialGRACES #TherapeuticAlliance #SocialWork