The UK Health and Safety Executive (HSE) describes culture as “the way we do things around here”: the context in which people judge whether their behaviour is appropriate. It treats safety culture as a subset of overall organisational culture. ^[1] The International Atomic Energy Agency (IAEA) defines it as the assembly of characteristics and attitudes in organisations and individuals which ensures that protection and safety issues receive the priority attention their significance warrants. ^[4]

The IAEA guide likens culture to an iceberg: above the surface are visible artefacts such as behaviour, words and symbols; below it are less visible elements such as norms, values and assumptions. ^[4] The sign at the site gate is only part of the surface.

Hence the difference between culture and climate. Safety climate usually refers to workers’ perceptions and attitudes at a given time, typically measured by survey. That is why HSE describes its Safety Climate Tool as measuring an element of safety culture, not the whole of it. ^[2]

Why don’t campaigns and slogans change the culture?

HSE identifies the largest influences on safety culture: management commitment and style, employee involvement, training and competence, communication, compliance with procedures, and organisational learning. ^[1] Nothing on that list is called an awareness campaign.

HSE also notes that many organisations focus on whether employees follow rules, when management culture matters even more, and it names a common mistake: confining change efforts to those below the manager who started them. It expects senior managers to accept that their own perceptions and behaviours will be examined and challenged. ^[1]

The same warning applies to behavioural safety programmes used as a substitute for working on culture. HSE sees them as a useful addition for occupational safety, but with limited benefits for controlling major hazards; they can shift responsibility from management to individuals, and they fail if trust is low or if workers see them as “another programme” or a blame culture. ^[3]

How do you measure safety culture?

The first rule in the IAEA’s self-assessment guide is that using multiple methods is essential, because each method gives different information and engages people differently. The guide recommends analysing each method’s results independently at first, to avoid premature conclusions. ^[4]

The second rule is that analysis starts descriptively: describe what you see and hear without mixing it with how things “should be”, and only then compare with the reference framework as a separate step. ^[4] If the assessment team arrives with a verdict already formed, it will find what it is looking for.

When bringing results together, look for commonalities and discrepancies across methods, choose representative examples while preserving anonymity, and go back to the data or collect more where a conclusion needs clarifying. ^[4] A contradiction between what the survey says and what observation shows is not a measurement error; it is often the most important finding.

In practice, a medium-sized organisation can start with a simple plan: a survey of all staff, separate focus groups for each job level so workers do not hold back in front of their supervisors, interviews with a number of supervisors and managers, observation of work across more than one shift, and a review of incident reports, procedures and safety meeting minutes.

The guide’s practical cautions include protecting respondents’ anonymity, because breaching it undermines any future assessment, and not relying on a single interpreter of the results, because every interpretation is biased to some degree. ^[4]

What does a safety climate survey ask?

If a questionnaire is one of your tools, borrow the structure of HSE’s Safety Climate Tool. It is an online survey of 40 questions with anonymous responses, measuring eight factors: ^[2]

• Organisational commitment.
• Health and safety oriented behaviours.
• Health and safety trust.
• Usability of procedures.
• Engagement in health and safety.
• Peer group attitude.
• Resources for health and safety.
• Accident and near miss reporting.

A useful feature of the tool is that it compares management and employee results and allows segmentation by group. ^[2] That comparison often matters more than the score itself: if management sees near-miss reporting as safe and workers see it as risky, you have found your first priority.

On Gulf work sites where people work in several languages, check that every group understands the questions and that responses are collected in a way workers trust will not reveal who they are.

Where does your organisation sit on the maturity ladder?

The Energy Institute’s Hearts and Minds programme presents a five-level ladder of safety culture maturity. ^[5]

The calculative level is a common trap for large organisations: lots of data and frequent audits, and a feeling of understanding how safety works, without always proving the data is effective. ^[5] At the proactive level, by contrast, the workforce takes an active role, line management leads on safety, and safety staff are mainly advisers. ^[5]

To locate your position on the ladder, the programme uses the Understanding Your Culture tool: a workshop of up to 20 people from across the organisation, each choosing the descriptions that best match how the organisation actually behaves, after which the results are presented and discussed by the group. An Arabic score sheet is available for version 5, which comes in several languages, while the latest version 6 is in English only. ^[6]

From measurement to change: what are the steps?

The IAEA guide sets out the self-assessment steps: prepare the organisation, prepare the team, prepare the plan, conduct the pre-launch, carry out the assessment, analyse, summarise the findings, communicate them, and develop and implement actions, with a follow-up within 6 to 18 months. ^[4]

What sets this path apart from a campaign:

1. Results go to management first, in a facilitated dialogue rather than an audit-style presentation; management then takes them to the rest of the organisation with the assessment team, mostly face to face, including shift workers and contractors. ^[4]
2. No quick fixes at the assessment stage: proposing immediate corrective actions misses the point of the assessment. ^[4]
3. The plan is built with the groups affected: a plan imposed from the top without real engagement does not build commitment. ^[4]
4. Focus on specific areas: HSE notes that targeting particular areas, such as safety leadership, competence or procedures, has produced positive results. ^[1]
5. Measure again: repeated measurement lets you track progress ^[1], and HSE’s Safety Climate Tool is designed to track improvement year on year. ^[2]

If you use an external provider, keep ownership of the process and build in-house skills so you can continue on your own, as HSE recommends. ^[1]

Hypothetical example: when good numbers are not enough

Hypothetical example: a maintenance company recorded a low injury rate, so its management assumed its safety culture was good. The survey showed a clear gap between management and workers on near-miss reporting, and interviews revealed that supervisors saw reports as evidence that their team was performing poorly.

The company did not launch a new campaign. It presented the results to senior management first, then agreed three actions with supervisors and workers: removing the number of reports from supervisors’ performance appraisals, responding to every report with what was done about it within a set time, and opening each weekly management meeting with one report and what was learned from it. It then planned to measure again after a year.

Conclusion

Safety culture is not measured by a poster or changed by a slogan. Measure it with more than one tool, describe before you judge, take the results to management before the workforce, build the plan with the people who will carry it out, and then measure again to see what has really changed.

Many self-assessments are full of phrases such as “the organisation is keen to”, “we apply best practice” and “we place great emphasis on”. These are claims an assessor cannot score, because they do not say what is actually applied, where, or what it achieved.

The problem is not just one of style. In EFQM’s recognition process, for example, the submission document is the assessors’ main working document, and the site visit then gathers evidence through interviews, focus groups and walkabouts. ^[2] Every sentence in the submission is a promise that will be tested.

How does an assessor read what you write?

The EFQM Model assesses organisations using RADAR logic. In the revised second edition of the Model (2021), Direction and Execution are assessed on three elements: approach (sound, with a clear rationale, responding to key stakeholders’ needs), deployment (implemented in relevant areas, effectively and on time) and assessment and refinement (feedback on effectiveness and efficiency is gathered, and trends, measurement and benchmarking drive improvement). ^[1]

The most important rule for the writer: no part of Direction or Execution can score higher than its Sound score. ^[1] If you do not explain why you chose the method and how it links to stakeholder needs, broad deployment will not rescue you.

For Execution specifically, two further attributes apply: the approach should be aligned, supporting the organisation’s direction and integrating with its other approaches, and deployment should be flexible, allowing adaptation to change. ^[1] So state how the approach connects with others, and how it was adjusted when circumstances changed.

EFQM’s chief executive describes the EFQM Model 2025 as the latest version and highlights RADAR logic as a core feature. ^[3] Still, check the attribute names in the guide for the edition you are assessed against.

The same logic in Gulf government models

Government models in the region have their own structures, so do not transfer one model’s criteria to another. The UAE Government Excellence Model, for instance, is presented on the official UAE government portal as GEM 2.1, with three pillars: distinctive value, vision realisation and enablers. ^[4]

The assessment logic, however, is similar. The Abu Dhabi Government Performance and Excellence Model (Version 2, 2026) scores a cycle of plan, implement, check and improve, followed by outcomes: meeting ambitious targets, sustained or improving performance, and favourable comparison with benchmarks. It uses a five-point scale from “None” to “All”, and states that the score for the plan’s foundation caps the scores for its alignment and implementation, while the score for suitable metrics caps the scores for data, improvement and outcomes. ^[5]

The practical lesson is the same in both: start with the foundation, meaning why and how the approach was designed, before writing about its spread and results. Keep a table that maps each criterion in the model you are assessed against to the corresponding approaches and evidence, so the same evidence is not repeated across criteria without adding anything new.

What is the difference between a claim and evidence?

The examples in the table are illustrative. The rule is that evidence answers “what, where, since when, how do we know, and what changed”.

Watch the verbs, too: claims use verbs of intention such as “strive”, “seek” and “emphasise”; evidence uses verbs of action such as “analysed”, “launched”, “reviewed” and “changed”. Changing the verb alone forces you to find the fact that proves it.

A template for evidencing each approach

Write each key approach as a short paragraph in this order, then add a clear reference to the evidence: the name of the document, system or report the assessor can ask for.

• What and why: name the approach as it is known inside the organisation, and state the need or analysis that prompted it.
• Where and since when: specify the departments or services covered, those not yet covered, and the start date.
• How we measure: the indicator, who reviews it and how often.
• What we learned: at least one change that came from a review or comparison.
• What we achieved: the result, with a cross-reference to the results section so figures are not repeated.

Government assessment guides point the same way. The leadership assessment guide issued by the UAE Prime Minister’s Office asks for submissions that are brief, prioritised and evidence-based with quantitative reference points, and encourages links or attachments supporting what is stated. ^[6]

Hypothetical example: rewriting a paragraph on service channels

Before: “The organisation is constantly keen to develop its digital channels in line with best practice, which has had a positive impact on customer satisfaction.”

After: “An analysis of service-centre visits in 2024 showed that most were enquiries about application status. We launched in-app application tracking for all main services, and review phone enquiries and visits monthly. After the first-quarter review we changed notifications to trigger at every status change. Status enquiries have fallen for three consecutive quarters (Service KPI Dashboard report, Appendix 4).”

The second paragraph contains the rationale, scope, measurement, refinement, result and evidence reference. The assessor can question the owner of each element in an interview.

How should results be written?

The EFQM Model assesses results on two elements: relevance and usability, meaning a clear link to purpose, vision and strategy and data that are timely, reliable, accurate and segmented enough to support improvement; and performance, meaning trends, targets, comparisons and future focus based on an understanding of cause and effect. Results cannot score higher than Scope & Relevance. ^[1]

So for every indicator in the submission:

• State why it matters and which strategic objective it serves.
• Show at least three years where available, not a single year.
• Put the target next to the actual and explain any gap.
• Add an external comparison where one exists, and name its source.
• Link the improvement to the approach that produced it.

Common mistakes in presenting results include percentages with no base number, indicators whose definitions change from year to year without explanation, aggregate results that are not segmented by service or group and so cannot guide improvement, and hiding indicators that declined instead of explaining the causes and the plan to address them.

Pre-submission checklist

1. Delete every sentence that starts with “we are keen to” or “we place emphasis on” and is not followed by evidence.
2. Check that every approach has a written rationale linking it to stakeholder needs.
3. Check that scope of deployment is specified: which departments or services, and since when.
4. Cite at least one improvement that resulted from a review, with its date.
5. Check that every figure in the results has a data source you can show.
6. Ask a colleague who did not write it to ask “how do we know?” of every paragraph.

Conclusion

A good self-assessment is not a marketing report about the organisation; it is a map of evidence the assessor can verify. For each approach, write its rationale, scope, measurement, refinement and result, and always start with the foundation, because its score caps what is built on it.