What is just culture?
Engineering culture and software practice
Just culture is an accountability model from aviation and healthcare safety science. People are not punished for honest mistakes or system-induced errors, so that they will report them, while reckless or wilful conduct is still held to account. James Reason described it as part of a safety culture, David Marx gave it a widely used three-behaviour model in 2001, and Sidney Dekker extended it. It is the foundation the blameless post-mortem stands on.
What this means
A just culture answers a hard question: when something goes wrong, how do you hold people accountable without making everyone too frightened to tell you the truth? Its core insight is that most people come to work to do a good job, and that punishing honest mistakes simply teaches staff to hide them, which makes the next failure more likely.
The model does not mean nobody is ever accountable. It draws a line. Honest error and system-induced slips are met with support and system fixes. Behaviour that knowingly and unjustifiably courted serious risk is still met with consequences. The skill is distinguishing the two fairly and consistently.
It has a name because "no blame" and "zap the person who erred" are both crude, and both fail. Just culture is the deliberate middle path, and it is the accountability model that a blameless post-mortem quietly relies on.
Why it matters
After an incident, human resources and engineering often pull in opposite directions. One instinct is to find who did it; the other is to understand how the system allowed it. A just culture gives leaders a shared language for both, so that accountability and learning are balanced rather than traded off.
The practical test many leaders remember is: what happens when someone deletes the production database? A just culture asks what choices that person made and what the system permitted, not merely how bad the result was. Whether people disclose the next mistake depends almost entirely on how the last person was treated.
This now extends to AI. If staff misuse an AI tool or feed it something they should not, they will only admit it in a culture where honest disclosure is safe. A blame-first response guarantees the organisation learns nothing until the problem is large.
Crucially, a just culture is not a shield against real duties. It distinguishes an honest mistake from at-risk behaviour and from recklessness, and it keeps consequences for the last of these while still learning from the first two.
How it works
Where the term came from
The idea grew out of aviation and healthcare safety science. James Reason, in Managing the Risks of Organizational Accidents (1997), described a just culture as one component of a safety culture, alongside a reporting culture and a learning culture. David Marx, an engineer and lawyer, transferred it into healthcare with his 2001 report "Patient Safety and the 'Just Culture': A Primer for Health Care Executives", written for the Trustees of Columbia University, which set out the widely used three-behaviour model. Sidney Dekker's Just Culture: Balancing Safety and Accountability (2007, with later editions) extended and critiqued the field, contrasting retributive and restorative versions.
The model spread into regulated practice. The International Civil Aviation Organization and Eurocontrol embedded it in aviation safety reporting, documented on SKYbrary, and Eurocontrol's definition protects operators for actions "commensurate with their experience and training" while not tolerating "gross negligence, wilful violations and destructive acts". In the United Kingdom, NHS England published "A just culture guide" (updated 14 December 2018), based on James Reason's incident decision tree. Software operations adopted the same thinking as blameless post-mortems became common practice.
The three-behaviour model
Marx's framework names three behaviours. Human error is inadvertently doing other than what should have been done: a slip, lapse or mistake, which calls for consoling the person and fixing the system. At-risk behaviour is a choice that increases risk where the risk is not recognised or is mistakenly believed to be justified, which calls for coaching and for removing the incentives that made the shortcut attractive. Reckless behaviour is a conscious disregard of a substantial and unjustifiable risk, which is rare and does warrant disciplinary action. The key move is that the response is tailored to the behaviour and circumstances, not to how bad the result happened to be.
The decision tests
Two tests help apply the model fairly. The substitution test asks whether a different, similarly trained and experienced person could plausibly have made the same error in the same situation; if yes, the finger points at the system rather than the individual. The foreseeability test asks whether the person could reasonably have foreseen the harm. Reason's incident decision tree, adapted in the NHS just culture guide, walks a manager through such questions to reduce the role of unconscious bias and treat people consistently.
Restorative versus retributive
Dekker distinguishes a retributive just culture, which asks what rule was broken and what the person deserves, from a restorative one, which asks who was hurt, what they need, and whose obligation it is to meet that need. Restorative approaches focus on repairing trust and preventing recurrence rather than on proportionate punishment. Most organisations blend the two, but naming the difference helps a leader decide what they are actually trying to achieve after an incident.
Running it in a small firm
You do not need a safety department. A small firm can adopt the three-behaviour distinction, agree in advance that honest error will be met with support, and make one senior person responsible for applying the tests consistently. The discipline is to react the same way whether the harm was trivial or severe, because staff watch how the last person was treated and calibrate their own honesty accordingly.
Examples
A retailer's warehouse system goes down after an engineer runs a maintenance script against the live database instead of the test copy. Under a just culture the manager applies the substitution test: the two environments looked almost identical and others had nearly done the same. The response is to fix the naming and add a confirmation step, and to support the engineer, rather than to discipline. Per a graded approach, discipline is reserved for behavioural choices, not for the severity of the result. This is an illustrative scenario.
A charity discovers a caseworker pasted client details into a public AI chatbot to draft a letter. The manager distinguishes at-risk behaviour, a choice made without recognising the risk, from recklessness. The proportionate response is coaching, a clear policy, and an approved tool, plus a look at why the sanctioned route was too slow to use. Punishing the disclosure would simply ensure the next such use is hidden.
A professional services firm's partner deletes a shared folder in anger after being told not to touch it. Here the tests point the other way: the risk was obvious, foreseeable and consciously disregarded. A just culture still holds this to account, which is what makes the protection for honest error credible to everyone else.
Common misunderstandings
The first misconception is that just culture means no consequences. Blame-free is not consequence-free; reckless conduct is still held to account, and the model exists precisely to make that line defensible.
The second is that it is the same as a blameless post-mortem. It is not. The blameless post-mortem is the meeting that reviews an incident and mentions just culture in passing, whereas just culture is the underlying accountability model and its decision rules, including where blame remains appropriate. The meeting is downstream of the model.
The third is that only front-line staff are accountable. Rather, managers carry their own accountability for the systems, incentives and pressures they create, and a genuine just culture applies the same scrutiny upward.
The fourth is that it removes legal or regulatory duties. It does not; disclosure, reporting and professional obligations still apply, and a just culture sits alongside them rather than replacing them.
The fifth is that it is a soft, HR-only idea. In aviation and healthcare it is a rigorous, decision-based framework with explicit tests, not a mood.
Risks and boundaries
The main risk is inconsistency. If managers apply the tests one way for a junior and another for a favourite, the culture collapses into cynicism faster than an openly punitive one. The model demands that the response track the behaviour, not the person or the severity of the harm, and that is genuinely hard to sustain.
There is also live debate. Dekker and others question whether organisations can ever draw the line between acceptable and culpable neutrally, since the person drawing it holds power. Critics in medical ethics argue that some framings understate legitimate demands for accountability to those harmed. And a just culture cannot override external legal duties: where the law or a regulator requires reporting or action, that stands. Leaders should treat the model as a tool for fairness and learning, not as a promise of immunity, and should take their own professional advice on legal obligations.
What to do next
Decide and write down, before the next incident, that honest mistakes will be met with support and system fixes. People need to know the rule in advance for it to change their behaviour.
Adopt the three-behaviour distinction and the substitution test as your default response to any incident, and make one senior person accountable for applying them consistently regardless of how bad the result looked.
Separate the learning conversation from any question of individual accountability, running them as distinct activities so that the review can be honest. The NHS just culture guide is a free, plain template you can adapt.
Watch your own accountability as a leader. If the incident traces to pressure, staffing or tooling you controlled, say so, because staff calibrate their honesty on whether accountability runs upward as well as down.
FAQs
Does a just culture mean nobody gets disciplined?
No. Honest error and at-risk behaviour are met with support and coaching, but reckless, wilful conduct is still held to account.
Who developed the idea?
James Reason described it, David Marx built the three-behaviour model for healthcare in 2001, and Sidney Dekker extended and critiqued it.
What is the substitution test?
It asks whether a similarly trained peer could have made the same error in the same situation; if so, the system, not the person, is the issue.
How is it different from a blameless post-mortem?
The post-mortem is the review meeting. Just culture is the accountability model, including its decision rules, that the meeting depends on.
Can a small business run a just culture?
Yes. It needs an agreed rule, the three-behaviour distinction, and one person applying it consistently, not a safety department.
Does it apply to AI misuse by staff?
Yes. It helps you separate an honest mistake from a risky choice from recklessness, and it makes disclosure safe.
Does it remove legal duties?
No. Reporting, regulatory and professional obligations still apply, and you should take your own advice on them.