Skip to main content
    QHSEtech
    Updated 4 October 2026
    Complete Guide

    Safety Culture Assessment Guide

    Safety culture cannot be reduced to one survey score or reporting count. This guide combines an attributed discussion framework with surveys, conversations, observations and operational evidence, so an assessment can identify questions and actions without pretending to certify an entire organisation.

    Editorial responsibility: Dimitris Mitsios (Founder of The QHSE Standard; product marketing at Tekmon) · Tekmon pays for sponsored placements on this site (disclosure) · How pages are made · LinkedIn · Content date: 4 October 2026

    TekmonAd

    Explore its published safety, quality and maintenance workflows. Confirm modules and pricing in a scoped proposal.

    Hearts and Minds/Hudson and Parker culture ladder

    The five labels below are the Hearts and Minds/Hudson and Parker safety culture ladder described by the Energy Institute. Use them as a facilitated discussion framework across multiple dimensions, not as a certification, a universal score or proof that an entire organisation occupies one stage. Sites, roles and topics can produce different patterns.

    1. Pathological

    The organisation gives little attention to safety beyond avoiding scrutiny or consequences.

    2. Reactive

    Safety receives attention mainly after an incident or other unwanted event.

    3. Calculative

    Systems, procedures and data are established, with safety managed through formal processes.

    4. Proactive

    The organisation looks for warning signs and involves people before an event occurs.

    5. Generative

    Safety considerations are integrated into ordinary decisions and learning across the organisation.

    Assessment Methods

    A questionnaire captures attitudes and perceptions at a point in time—an element of culture, not the whole of it. Combine it with worker discussions or interviews, observations, incident and near-miss learning, inspections and audits, plus evidence about management-system and engineering controls.

    Safety Perception Surveys

    Anonymous questionnaires measuring worker perceptions of management commitment, safety communication, reporting culture, and personal responsibility.

    Behavioral Observations

    Structured observation programs where trained observers record safe and at-risk behaviors during routine work activities.

    Leading Indicator Analysis

    Use rates with denominators, timeliness and qualitative follow-up. A rise or fall in near-miss reports alone does not prove improvement: access, trust, scope and under- or over-reporting can change the count.

    Management Commitment Audit

    Evaluate leadership behaviors: safety walk frequency, resource allocation, incident response quality, and visible safety engagement.

    Tools to Support Cultural Change

    Software such as BBS platforms, observation tools and near-miss reporting apps can support confidential collection, segmentation, action ownership and feedback. It does not prove cultural improvement. Review response coverage, action quality and other evidence alongside QHSE indicators.

    Find Safety Culture Tools

    How an assessment cycle runs

    The figure shows one workable order and the record each step should leave. The purpose comes first because it decides which methods are worth the effort, and the discussion step sits before actions because the workforce can say what a number means in practice.

    Safety culture assessment cycleSeven steps across four roles, from agreeing the purpose to repeating the assessment with a comparable method, ending in a loop back to the start of collection.SponsorAssessment leadAnalystManagers and workersNext cycleStep 1Agree the purposeand scopeThe decision theresults will inform,and who is in scope.Step 2Choose methods anda baselineSurvey, discussions,observations andrecords to becombined.Step 3Collect responsesand evidenceResponse coverage bygroup, withconfidentiality rulesstated.Step 4Analyse by groupand topicSmall groups withheld;differences betweensites noted.Step 5Discuss resultswith the workforceWhat the numbers meanin their work, andwhat is missing.Step 6Agree actions withownersNamed owner, date andthe evidence ofcompletion.Step 7Repeat with acomparable methodSame questions andgroups, after actionshave had time to work.
    Names and the number of steps vary by organisation. The assessment does not end when the report is written; the actions and the repeat are part of it.
    Text version of this diagram
    1. Step 1. Agree the purpose and scope. Owner: Sponsor. The decision the results will inform, and who is in scope. Then step 2.
    2. Step 2. Choose methods and a baseline. Owner: Assessment lead. Survey, discussions, observations and records to be combined. Then step 3.
    3. Step 3. Collect responses and evidence. Owner: Assessment lead. Response coverage by group, with confidentiality rules stated. Then step 4.
    4. Step 4. Analyse by group and topic. Owner: Analyst. Small groups withheld; differences between sites noted. Then step 5.
    5. Step 5. Discuss results with the workforce. Owner: Managers and workers. What the numbers mean in their work, and what is missing. Then step 6.
    6. Step 6. Agree actions with owners. Owner: Sponsor. Named owner, date and the evidence of completion. Then step 7.
    7. Step 7. Repeat with a comparable method. Owner: Assessment lead. Same questions and groups, after actions have had time to work. End of the sequence.

    Feedback loops

    • From step 7 (Repeat with a comparable method) back to step 3 (Collect responses and evidence): Next cycle.

    Questions to settle before you collect anything

    • What decision will the results inform, and who will see them? Say so before anyone answers.
    • Which groups will be reported separately, and what is the smallest group you will show? Set the minimum before the data arrives.
    • Who has access to individual responses, and how are comments handled when they could identify a person or a team?
    • How will responses reach people who do not use a desk, a shared computer or the intranet?
    • What will you do if the results are unwelcome? Agree this with the sponsor in advance.
    • When will findings and actions be shared with the people who answered, and by whom?

    The HSE Safety Climate Tool help guidance, opened on 4 October 2026, says organisations should assure anonymity or confidentiality of responses, says its automated report is produced once a minimum of 10 responses have been collected, and advises communicating the results, acting on the findings and monitoring the action plan. The tool's own page speaks of discussing results with workers and tracking improvements year on year. Use it as a worked example of how one public tool handles these points; your own minimum group size is a decision for you to take and record.

    Reading survey figures: a worked example

    The figures below are invented to show the arithmetic and are not a benchmark. Suppose 400 people are invited and 142 respond.

    • Response rate: 142 ÷ 400 = 35.5%. The rate says who is represented, not how they feel, and it needs to be read by group because one site may be at 60% and another at 15%.
    • One group has 48 respondents, and 31 of them chose the two most favourable points on a question: 31 ÷ 48 = 64.6% favourable.
    • In this group one respondent moves the figure by about 2.1 percentage points (1 ÷ 48).
    • A repeat survey the next year, with the same question and group, has 33 of 48 favourable: 68.8%. That is 4.2 points higher, which is two respondents.

    A difference of two respondents is small enough that a different mix of people answering, a changed question or an event in the week of the survey could account for it. The rise is something to ask the group about, not evidence that the culture changed. Where the same people answer twice and the survey is anonymous, you cannot tell whether the same individuals moved.

    Common mistakes

    Reporting one score for the whole organisation

    An average hides differences between sites, shifts, contractors and roles. Report by group, with the group size, and look first at where the answers differ.

    Treating a good result as proof

    A favourable survey shows what people said they perceive. It does not show that controls work. Compare it with inspections, investigations, observations and the quality of actions taken.

    Using results to judge individuals or teams

    If people expect a score to be used against a supervisor, answers can change. State in writing what the data will and will not be used for.

    Surveying without acting

    A second survey that finds the same concerns, with no visible action in between, can lower trust in the process. Share what was heard and what will change before repeating.

    Changing the questions every time

    Year-on-year comparison needs the same questions, scale and groups. If a question must change, show the old and new results separately.

    Reading reporting counts as culture

    A rise in near-miss reports may mean more people feel able to report, or that a new form is easier. Either way the count alone does not tell you which, so read it with the discussions behind it.

    Sources and interpretation limits

    The ladder is a discussion aid, and a safety-climate survey measures worker attitudes and perceptions rather than the whole culture. ISO 45004 describes combining qualitative and quantitative information and warns that reporting behaviour can distort indicators. Select methods and review intervals for the assessment purpose and action plan.

    Related Guides

    QHSE KPIs & MetricsBehavior-Based Safety SoftwareTraining Management SoftwareBest Safety Management SoftwareNear Miss Reporting Software

    Frequently Asked Questions

    What is a safety culture assessment?
    A safety culture assessment evaluates the shared beliefs, values, attitudes, and behaviors regarding safety within an organization. It goes beyond compliance metrics to measure how safety is perceived and practiced at every level — from executives to frontline workers.
    How do you measure safety culture?
    Common methods include: safety perception surveys (anonymous questionnaires), safety climate audits, behavioral observation programs, leading indicator analysis (near-miss reporting rates, safety meeting attendance), and management commitment evaluations.
    How often should safety culture be assessed?
    Set the interval from the assessment purpose and action plan. Establish a baseline and repeat with a comparable method after actions have had time to take effect, or earlier after a material organisational or risk change or evidence that the plan is not working. An annual cycle may suit some programmes, but it is not a universal requirement.
    What response rate do I need?
    No universal figure is established here. What matters is who is represented: a high overall rate can still leave one site, shift or contractor group silent. Report the rate by group, say which groups are thin, and avoid drawing conclusions about a group with few answers.
    Should the survey be anonymous?
    Confidentiality is a common aim, and the HSE Safety Climate Tool guidance stresses assuring respondents of it. Decide what you can honestly promise, including how free-text comments are handled, and state it in the invitation. If you promise anonymity you cannot then ask for details that identify a person.
    Is the culture ladder a score?
    No. It is a facilitated discussion framework from the Energy Institute's Hearts and Minds material, and different sites, roles and topics can sit at different stages. Use it to start a conversation about evidence, not to place the whole organisation at one level.
    Can software improve safety culture?
    Software can support anonymous collection, segmentation, action ownership and feedback. It does not itself establish culture improvement. Assess response coverage, confidentiality, action quality and evidence from discussions, observations, investigations, inspections and audits alongside the counts.