Form questions
Safety culture survey questions for a site or plant
Twenty four questions suits an annual round and ten suits a quarterly one. Three cannot be cut: whether a worker can stop a job they think is unsafe without arguing for it, how many near misses they saw against how many were reported, and one named hazard that has not been dealt with. Everything else is context for those three.
A safety culture survey measures whether people believe it is safe to speak, which is a different question from whether the paperwork is in order. These questions ask about stopping a job, reporting a near miss and naming a hazard, on an Australian site where the crews answer anonymously.
The questions
Your site
Which site do you work on most?DropdownRequired
Include a prefer not to say option. Site is the only identifier this survey should hold.
How long have you worked here?Dropdown
Optional, with prefer not to say. New starters and twenty year veterans answer the reporting questions very differently.
What kind of work do you mostly do?Dropdown
Optional. Trade, supervisor, plant operator, apprentice, office, with prefer not to say.
Do you work alone for part of a normal shift?Single choice
Lone workers score differently on almost every row below.
Reporting and speaking up
I can stop a job I think is unsafe without arguing for itRating gridRequired
Row one of a five point agreement grid. The single most useful row in the survey.
Reporting a near miss is treated as useful, not as a complaintRating gridRequired
Row two.
My supervisor acts on hazards within the weekRating gridRequired
Row three. A named timeframe makes the row answerable.
People here say something when they see a shortcutRating gridRequired
Row four.
I would report a mate doing something unsafeRating gridRequired
Row five. The row that separates a compliant site from a safe one.
How many near misses did you see in the last month?Number
An estimate is fine. The number matters less than the comparison with the next question.
How many of those were reported?Single choice
All, most, about half, a few, none, or I saw none. The gap is the under reporting rate.
If you did not report something, what stopped you?Multiple choice
Too busy, the paperwork, nothing would happen, it would get someone in trouble, I did not think it counted.
When did a supervisor last talk to you about safety on the job?Single choice
Not a toolbox talk. This week, this month, this year, never, cannot remember.
Equipment and plant
Is the right protective equipment available when you need it?Single choiceRequired
Always, usually, sometimes, rarely. The only equipment question that must be answered.
What is hardest to get hold of on site?Multiple choice
List real items for the trade. Include a nothing is hard to get option.
Is plant and equipment maintained well enough to use safely?Single choice
Have you used equipment you knew was not right for the task?Single choice
Never, once, a few times, regularly, prefer not to say. Expect the honest answers only under anonymity.
How often do you feel rushed enough to skip a step?Single choice
This is the production pressure question. Ask it about frequency, not about agreement.
Training and readiness
Were you shown how to do this job safely before you started it?Single choice
Do you know what to do if someone is injured on your site?Single choice
Yes and I have done it, yes in theory, not sure, no.
Is the safety training you receive relevant to the work you actually do?Single choice
How useful are the toolbox talks?Single choice
Include a do not attend them option so the average is not diluted.
What would you fix
Name one hazard on your site that has not been dealt withLong textRequired
Ask for what and where, and say plainly that nobody should be named.
What would make it easier to raise a safety issue here?Long text
Which questions to make required
This survey is anonymous by default. There is no name field, no employee number and no email question anywhere in it, and both demographic questions offer prefer not to say. Only four things are required: the site, the five grid rows, protective equipment availability and the named hazard. The site is required because a result that cannot be attributed to a location cannot be acted on, and the grid rows are required because a partly completed grid cannot be averaged. Everything about near misses, pressure and training stays optional.
- Which site do you work on most?
- I can stop a job I think is unsafe without arguing for it
- Reporting a near miss is treated as useful, not as a complaint
- My supervisor acts on hazards within the week
- People here say something when they see a shortcut
- I would report a mate doing something unsafe
- Is the right protective equipment available when you need it?
- Name one hazard on your site that has not been dealt with
The form built from these questions
A static copy of a form built from part of the bank, exactly as it renders.
How to change this for a different kind of workplace
An office or a professional workplace. Physical hazards are rare and psychosocial ones are not, so replace the equipment section with questions about workload, unreasonable deadlines, customer aggression and whether people know how to report bullying. Keep the whole reporting grid unchanged, because the willingness to speak up is the same construct in both settings. A construction site with subcontractors. Add a question about who the respondent actually works for, since a subcontractor on a principal contractor site often has two safety systems and trusts neither. Add a row to the grid about whether the site induction matched the real work. Expect the subcontractor answers to be worse and resist the urge to report them separately if the crews are small enough to identify. Healthcare or aged care. Replace protective equipment with questions about staffing levels on the hard shifts, manual handling equipment and occupational violence, and add a question about whether incidents involving patients or residents are reported the same way as incidents involving staff. In most services they are not, and that is the finding.
Questions people ask
Why must a safety culture survey be anonymous?
Because the thing being measured is whether people feel safe speaking. A survey that records who said what measures something else entirely, and it reliably returns reassuring numbers from a workforce that has learned what a safe answer looks like. Anonymity is not a courtesy here, it is the instrument working.
Is this survey a hazard report?
No, and the form should say so on both the first and last screen. A hazard report goes through the reporting system and is acted on immediately by someone whose job it is. A periodic culture survey may sit unread for a fortnight, so anything urgent has to be sent down the other path.
How small a crew can be reported on its own?
Eight responses is a reasonable floor for a trade workforce, and it should be printed where the crews can read it. Below that, fold the site into the company total and say that you have done so. A site of five people reported separately is not anonymous no matter what the cover note claims.
How often should it run?
Annually for the full instrument, with a short pulse of five or six questions after a serious incident or a major change to the work. Running it monthly produces survey fatigue and, worse, an expectation of action that a safety committee meeting quarterly cannot meet.
What does a good result look like?
On the stop work row, anything below eighty per cent agreeing is worth investigating. The near miss comparison is more diagnostic than any average: a site reporting twenty observed near misses and two reported ones has a reporting problem regardless of how well it scores on everything else.
Should supervisors answer the same survey?
Yes, with a role question so the two groups can be compared. Supervisors almost always rate the culture higher than their crews do, and the size of that gap is a better predictor of trouble than either group score on its own.
How would this change for driving work?
Fatigue becomes the main hazard, so add questions about roster notice, scheduling pressure and whether drivers feel able to stop when tired, asked as frequency questions rather than as statements to agree with. Keep the stop work row, because on the road it is the same question asked about a different kind of job.
Build this form in about a minute
The button opens the form generator with this form already described. Change the wording to match your own, generate, then edit any field.
Create a form with OneCraftOther question banks
Employee wellbeing survey questions that are not a diagnosis
A workplace wellbeing survey should measure the conditions an employer controls, not the state of a person. These questions ask about workload, rest, recovery and whether people know what support exists, and none of them is required.
Meeting feedback questions that take under a minute
Meeting feedback only gets answered if it costs less than the meeting did, which means most of it has to be one tap. These questions cover purpose, attendance, decisions and length, and they end with the only question that changes a calendar.
Benefits survey questions that rank what matters
A benefits survey that asks what people would like collects a list of everything, priced at more than the budget. These questions ask what was used, what would be kept if the budget shrank, and what would be given up to fund something better.
On writing the questions themselves, read How to write survey questions. For everything the builder can do, see the form maker. Step by step: Control who can fill in your form, then Create a form with AI.
Written and checked by the OneCraft team. Last checked .