Behavioural Safety: 4 Questions Before You Buy a Programme
Somebody has recommended a behavioural safety programme. It comes with observation cards, peer-to-peer feedback training, a dashboard of leading indicators and a number you can report to the board. It sounds like maturity, and it arrives at exactly the moment an organisation has run out of obvious things to fix.
Before signing, it is worth reading what the regulator says, because it is markedly more cautious than the marketing. The HSE's own human factors material sets out the model fairly and then lists its failure modes: these programmes tend to focus on easy and visible issues while overlooking low-probability, high-consequence risks, they can shift responsibility from management onto individuals, and they may not address the effects of management behaviour at all. Its conclusion is measured rather than hostile. Behavioural safety is a useful addition to the toolkit, particularly for occupational safety, and of limited value for controlling major hazards.
That is a reasonable verdict for a factory. For a law firm, an insurer, a university or a software business it raises a harder question, which the HSE also asks: whether behaviour change is the right fix at all, compared with a better procedure or equipment that is easier to use.
Here are four questions worth answering before you buy, each taken from the regulator's own cautions rather than ours.
What behavioural safety actually is
Worth stating plainly, because the term gets used for anything involving people.
The behavioural safety model has three steps. Define which behaviours are safe and which are unsafe. Observe work as it happens, by managers or by peers. Then give feedback that reinforces the safe behaviours and re-educates the unsafe ones. Feedback ranges from an immediate conversation on the spot to anonymised data reported in aggregate.
The approach rests on the premise that many accidents are attributed to inappropriate behaviour, and the claimed benefits are real enough: visible management presence, employee engagement, better safety conversations and a set of leading indicators rather than a lagging injury count.
Everything that follows assumes the model works as described. The questions are about whether it fits what you have.
1. Is behaviour the controlling factor, or is the fix a better design?
The first question, and the one the HSE puts most directly. It asks whether behaviour change is the right intervention compared with better procedures or equipment that is easier to use.
This matters because an observation programme will always find behaviour, since behaviour is what it looks for. A person reaching across a desk to a badly placed socket is behaving unsafely and is also responding rationally to a socket in the wrong place. Move the socket and the behaviour disappears without a single conversation.
The honest test is to take your last ten incidents and ask, for each, whether the person could reasonably have behaved otherwise given the equipment, layout, workload and procedure in front of them. Where the answer is no, a behavioural programme will record the behaviour indefinitely and never fix the cause. Our note on incident investigation covers how to get past the behaviour to the condition underneath.
The HSE frames this as a question about rules as much as people: whether you have a rule violation or a bad rule.
2. In a professional environment, what is there to observe?
Need Expert H&S Guidance?
Our qualified consultants can help you implement the right health & safety measures for your business.
The practical question, and the reason most office-based programmes quietly stop after a year.
Peer observation was developed where work is physical, visible, repeated and performed in shared space. An observer can watch a task, compare it against a defined safe method and have a conversation about the difference. That is a sound design in manufacturing, process industries or rail.
In a professional firm the work is cognitive and the behaviours that matter are mostly invisible. There is no observable moment that distinguishes a well-managed workload from an unmanageable one, a person who has read the evacuation procedure from one who has not, or a lone worker who told somebody where they were from one who did not. What an observer in an office can see is posture, cables, propped fire doors and boxes in corridors, which is a short list that an inspection already covers.
So the sequence to be honest about is this: if the observable items are already covered by inspection, and the consequential risks are not observable, the programme is measuring the wrong things with considerable effort.
3. Will it reach management behaviour, or only individual behaviour?
The question the HSE raises most pointedly, and the one that decides whether a programme builds trust or destroys it.
The caution is that these programmes can shift responsibility from management onto individuals, and may not address the effects of management behaviour. In a professional services firm that risk is not theoretical, because the dominant occupational health risk is workload and deadline pressure, and those are produced by management decisions about resourcing, utilisation targets and client commitments.
An observation programme pointed at individuals in that environment produces a predictable result. It records that people worked late, skipped breaks and sat badly, attributes those to personal choice, and leaves untouched the decisions that made them the rational response. Staff read that accurately and quickly, which is why the engagement benefit reverses.
If a programme cannot generate findings about management behaviour, resourcing and job design, it is not a safety culture programme. Our note on work-related stress sets out the duties that sit on the organisational side of that line.
4. Will it chase the visible and miss the consequential?
The last question, and the one that matters most to anyone with a genuinely serious low-frequency risk.
The HSE's warning is that behavioural programmes gravitate to easy, visible issues and can overlook low-probability, high-consequence risks. Its verdict on major hazards is explicit: limited value.
Translate that into our sectors. A university's most serious risks are a laboratory incident, a fire in a hall of residence and a legionella outbreak in a building nobody occupied all summer. A technology business with a data hall has electrical and suppression risks. An insurer in a multi-let tower has evacuation of people who need assistance. None of these is a behaviour you can observe on a Tuesday, and all of them are low-frequency and high-consequence, which is precisely the category the HSE says this approach handles poorly.
A programme that reports 400 observations a month and has nothing to say about any of the above is producing activity rather than assurance.
The four questions, and what a yes or no means
| Question | If the answer is unhelpful | What to do instead |
|---|---|---|
| Is behaviour the controlling factor? | Incidents trace to layout, equipment or workload | Fix the condition. Behaviour follows the design |
| What is there to observe? | The observable items are already on an inspection | Strengthen inspection, drop the observation layer |
| Will it reach management behaviour? | Findings can only be about individuals | Redesign it to generate findings about resourcing and job design |
| Will it reach the consequential risks? | It cannot touch your low-frequency, high-consequence risks | Keep those in a separate assurance route and do not let volume substitute for it |
Four unhelpful answers does not mean doing nothing. It means the intervention is the wrong shape, and the money is better spent on the conditions and on the assurance route for the risks that would actually hurt somebody.
What does transfer
Three elements of the model work well in professional environments, which is worth saying because the critique above is not an argument for inertia.
Defining what good looks like is useful everywhere, provided the definitions describe conditions and arrangements rather than personal virtue. Feedback loops are useful, and the strongest version in an office is near miss reporting with a visible response, which our note on near miss reporting covers. And leading indicators are genuinely better than injury counts, as long as the indicators measure the organisation rather than the individual: assessments reviewed on time, actions closed within target, training currency, evacuation arrangements tested.
That is the distinction worth holding. Measure the system's behaviour, not the staff's. Our note on safety culture sets out which measures survive contact with a professional firm, and the HSE's framework in HSG65 and its guidance on managing for health and safety put behaviour where it belongs, inside a management system rather than beside one.
Holding those indicators in the same place as the assessments and actions they refer to is where health and safety consultants and software are worth more in combination than either alone, because a leading indicator calculated by hand each quarter stops being calculated. Periodic health and safety audits should then ask what changed as a result of the programme, which is a question that separates the two kinds very quickly.
Why the international picture is moving away from observation
For multi-country groups there is a further reason to be careful about buying a behaviour-led programme as a global standard: the direction of regulation is the other way.
Several jurisdictions now require management of organisational factors by law rather than as good practice. Brazil has brought work-related psychosocial risk factors expressly inside the occupational risk management programme its labour ministry approves, enforceable since May 2026, so workload, working hours, harassment and unclear roles must be identified, assessed, classified, controlled and monitored like any other risk. Australia has written psychosocial hazards into its work health and safety regulations with a code of practice behind them. Mexico has required identification and analysis of psychosocial risk factors for several years.
Each of those targets the thing behavioural observation does not reach. A group investing in the observation of individuals, in a regulatory environment moving towards duties about job design and workload, has bought a programme aimed at the previous decade's question. The structural answer holds everywhere: one assessment methodology covering organisational factors as well as physical ones, with the local instrument and interval attached per country.
Where Arinite fits
Arinite has spent 15+ years in organisations where the serious risks are infrequent, the work is cognitive and the dominant health risk is produced upstream of the person doing the job. That is a different problem from the one behavioural safety was designed to solve, and saying so before taking a brief has cost us work and saved clients more.
We support 1,500+ businesses in 50+ countries and protect 100,000+ employees, with 95%+ client retention. Our health and safety consultants work extensively with legal, insurance, education and IT and software clients, and where a behavioural element genuinely helps we build it into health and safety training rather than running it as a parallel programme with its own dashboard.
For multi-country groups, our global health and safety consultants build the organisational factors into the local instrument each country requires, and our international health and safety consultants can tell you which entities now have a legal duty your current culture programme does not address.
Take your last ten incidents and ask, for each one, whether the person could reasonably have acted differently given what was in front of them. Then book a free gap analysis and we will tell you in one call whether you need a behavioural programme or a better set of conditions.
Related Articles
Written by
Arinite Health & Safety Consultants
Health & Safety Expert at Arinite
