The argument in brief
Two claims, honestly made
Culture is an inherent risk. Public risk guidance names it as an internal driver, and the evidence shows exclusionary cultures silencing the knowledge, warnings and unheard voices an organisation needs to detect its own hazards.
Inclusion is the control, not a values programme. Done as rhetoric it can misfire or quietly preserve the status quo. Done as designed, resourced, owned controls, preventative, detective and directive, it narrows the distribution of outcomes: the catastrophic tail becomes less likely, and the voices that would have warned of it are finally on the register, with an owner, a rating and a review date.
The reframe
Two questions, not one
Most arguments about culture quietly run two different questions together. They do not have the same answer. Select each to see what the evidence actually supports.
The CIPD and CEBMa evidence review found little consistent link between culture and performance, a mean correlation of approximately 0.16 as reported, and no valid, reliable evidence that culture-change programmes improve performance. If the case for managing culture rested here, it would deflate.
Risk is about downside variance and tail outcomes, not mean uplift. Toxic culture was the single best reported predictor of Great Resignation attrition, roughly ten times more powerful than pay, and the non-inclusive cluster was its strongest component. Inquiries from Grenfell onward show what happens when the warning voices go unheard. This is the question this paper answers.
Meeting the sceptical evidence
The numbers, as reported
These findings do not conflict. Attrition, safety and voice are the downside channel; profit is the mean-uplift channel. The honest claim is the narrower one: exclusionary culture is a measurable source of specific risks, and those risks can be controlled. Figures are reproduced exactly as their sources report them; full caveats and citations sit in the paper's endnotes and evidence register.
The mechanism
How exclusion crystallises into risk
The silencing of knowledge
High-context communication carries meaning in relationships and setting; low-context processes demand everything be stated, written and formalised. Build your processes on low-context assumptions and you systematically exclude people who operate in high-context modes, not through prejudice but through cultural misalignment (Hall, 1976; Piwowar-Sulej, 2020). On a safety-critical site, that is not an inconvenience. It is a hazard. A two-wave study of 203 employees found perceived diversity feeding interpersonal conflict through negative affect, and inclusive practices, top-down and bottom-up, weakening the effect (Liu, Zhu and Wang, 2023). Unmanaged difference is the hazard; inclusion is the moderating control.
The absence of psychological safety
A culture that cannot hear difference cannot surface its own hazards, because people who do not feel safe do not report what they see. The Grenfell Tower Inquiry's Phase 2 report (2024) found residents had repeatedly raised fire-safety concerns that went unheeded, before a catastrophe the Inquiry judged avoidable. The built environment's workforce data shares the root: the highest fatal-injury count of any Great Britain sector, markedly elevated suicide risk among male construction workers, and among the widest gender pay gaps (HSE; ONS). A sector that silences loses its early signals.
The loop that resists correction
Personal bias feeds organisational systems, which perpetuate societal bias, which reinforces personal bias: a reinforcing feedback loop (Meadows, 2008; the Unhelpful Bias Cycle, Ramroop, 2024). Stroh (2015) names why it resists repair: the hidden benefits of business as usual, predictability, in-group cohesion, low-friction promotion. Until those quiet rewards are named, the system compensates against every change effort. Good intentions are not enough.
The machinery beneath: fragmentation and requisite variety
Aggregate culture scores average away exactly the fragment where harm lives. In the toxic-culture data, no diversity topic predicted a company's overall rating, yet the non-inclusive cluster was the strongest predictor of whether an individual experienced the culture as toxic (Sull et al., 2022). Soane (2025) shows why: subcultural fragmentation is itself a source of risk, hiding pockets of bullying and practical drift that integration-focused measures overlook. And Ashby's requisite variety principle, applied to cultural risk by Firsova (2026), names the deepest layer: an organisation facing a varied world needs matching internal variety to read it. Variety absorbs variety. A homogeneous organisation is under-equipped to sense the hazards its environment generates.
The capability
Cultural Intelligence, in full
Cultural Intelligence, CQ, is the capability to function and relate effectively across difference: nationality and faith, and equally gender, generation, class, disability, neurodivergence, profession. It is not a trait and not a synonym for awareness. The construct originates with Earley and Ang (2003), was validated through the research programme collected by Ang and Van Dyne (2008), and is applied worldwide through the Cultural Intelligence Center's four-capability model, of which CQ is a registered framework. In Building Inclusion (Routledge, 2024) we apply it systemically, as the operating discipline of inclusive culture change.
CQ Drive
Whether you actually want to work and relate across difference: interest, motivation and the confidence to manage unfamiliar situations. In short: do you mean it?
CQ Knowledge
What you know about how values, norms, communication styles and lived experiences differ, held as patterns to inform curiosity, never as stereotypes to apply to individuals.
CQ Strategy
Planning, awareness, checking. The discipline of pausing to question your assumptions before you act. Without it, Drive plus Knowledge curdles into tokenism. With it, decisions become conscious.
CQ Action
The adaptations you actually make, in speech, non-verbal behaviour and flexibility, so that what you do is appropriate to the context and authentic to you.
Three properties make CQ a risk capability rather than a virtue: it is measurable, through validated assessment, so it can be baselined and reviewed like any control; it is developable, so investing in it is rational control spend; and it cascades, from leaders into policies, processes and culture. None of it holds without resource: STREAM, Support, Time, Resource, Effort and expertise, Agency, Money, is the entry condition for every control on the register below. A control with no budget and no authority is a policy on paper.
The signature
Watch the register move
Ten organisational-culture risks, plotted by likelihood and impact. Apply the CQ controls and watch what actually changes: likelihood falls, while severity often holds, because a fatality or an unheard community remains catastrophic if it occurs. Inclusion narrows the distribution. It does not promise elimination. Select any risk for its full register entry.
Impact → (vertical axis: Likelihood ↑)
Select a risk on the map
Each numbered circle is a register entry. Choose one to see its cultural mechanism, its CQ mitigations mapped to control classes, its statutory hook, its owner and its STREAM resourcing.
Deliberately included are the two entries most registers omit: 09, inclusion done badly, a risk in its own right (initiatives designed without the people affected can harm the very people they target, Cassell, Watson, Ford and Kele, 2022), and 10, compliance theatre, statutory duties held without resource. Scores are illustrative; recalibrate to your organisation's risk appetite. The full register, with live scoring, accompanies the paper as a working spreadsheet.
Statutory levers
The law is already carrying the argument
UK law increasingly frames culture management in this paper's terms: anticipatory duty, risk assessment, reasonable steps, the surfacing of unheard concern. Positions stated as at July 2026; check current sources before reliance.
- October 2024
Worker Protection (Amendment of Equality Act 2010) Act 2023 in force: a proactive duty to take reasonable steps to prevent sexual harassment. Harassment becomes an anticipatory risk-management obligation, evidenced by documented assessment. - February 2025
Procurement Act 2023 main provisions in force: inclusion and social value can be embedded as award criteria, giving inclusive supply-chain choices a legislative basis. - April 2026
Employment Rights Act 2025 begins phasing in: sexual harassment becomes a qualifying protected disclosure, and Equality Action Plans open on a voluntary basis. Alongside these sit the standing duties: the Health and Safety at Work etc. Act 1974 and 1999 Regulations reaching psychosocial risk, and the Public Sector Equality Duty. - October 2026
The harassment standard rises to all reasonable steps, and direct employer liability for third-party harassment arrives across all protected characteristics. - 2027
Equality Action Plans become mandatory for larger employers; defining regulations on the all-reasonable-steps standard expected. - In progressNot yet law
The Public Office (Accountability) Bill, the Hillsborough Law, would create a statutory duty of candour for public authorities, drawing explicitly on Hillsborough, Horizon, Grenfell and infected blood. Carried over to the session beginning May 2026. A clear signal of direction, not yet binding.
A statutory duty creates the obligation; it does not by itself reduce the residual risk. The statutes are directive controls. Whether they function as real controls or compliance theatre is determined by culture, and by Cultural Intelligence as its operating capability.
There are no quick fixes, no magic wands, no silver bullets. Just conscious, inclusive actions.
Put the unheard voice on the register, with an owner, a rating and a review date. Then act.