Access a quick summary of what a safety observation is with AI.
A safety observation is a record of a workplace condition or behaviour that could affect someone’s safety – whether it’s something going wrong or something being done right. Unlike an incident report, it’s made before anyone gets hurt, which makes it one of the clearest leading indicators a safety team has.
What is a safety observation?
A safety observation is a documented note of what someone saw at work that has a bearing on safety. It might be a trailing cable in a walkway, a colleague skipping a pre-use check, or a team running a lift exactly the way the plan said they should.
The defining feature is timing. A safety observation is typically submitted before an incident occurs – while there is time to do something about it, before harm occurs. This is why observations sit alongside near misses at the top of most proactive safety programmes.
Two things separate a useful observation from a box-ticking exercise – it names something specific, and it gets closed out. An observation that sits in a spreadsheet with no owner and no action is a data point, not a control.
Safety observation vs near miss vs hazard vs inspection
These four terms get used interchangeably in day-to-day conversation, and that costs teams accuracy in their reporting. Here is how they differ.
| Term | What it records | Trigger | Timing |
| Safety observation | A condition or behaviour - unsafe or safe | Someone notices something | Anytime |
| Near miss | An event that could have caused harm but didn't | A specific event happened | Immediately after the event |
| Hazard | Something with the potential to cause harm | Identification of a source of risk | Anytime, often formalised in a risk assessment |
| Inspection | A planned check against a defined standard | A schedule | Scheduled |
| Audit | Whether the safety management system works | A schedule | Periodic |
The practical distinction is that a near miss needs an event; an observation does not. If a pallet falls and misses someone, that’s a near miss. If you walk past badly stacked pallets and report them, that’s a safety observation. Report the observation, and you may never have to report the near miss.
The two types of safety observation
Unsafe observations
These record a condition or behaviour that increases risk – a missing guard, a blocked fire exit, someone working at height without a harness. They’re what most people picture when they hear ‘safety observation,’ and they generate the corrective actions.
Positive (safe) observations
These record someone doing the right thing – stopping a job because conditions changed, correcting a colleague, using the right equipment when nobody was watching.
Often, they are underused. Positive observations do two jobs at once. They tell you which controls are actually working in practice rather than on paper, and they change how reporting feels on the floor – from a system that catches people out to one that recognises good work. Product Care Group recorded a 200% increase in positive observations after moving to Notify, and the reporting culture shifted with it.
A healthy programme records both unsafe and safe observations.
15 examples of safety observations
| # | Category | Example |
| 1 | PPE | Operative working in the yard without hi-vis or safety footwear |
| 2 | PPE | Correctly fitted RPE with an in-date face-fit certificate |
| 3 | Housekeeping | Cables run across a walkway without matting |
| 4 | Housekeeping | Spill kit used immediately after a leak, area cordoned |
| 5 | Access & egress | Fire exit blocked by stacked stock |
| 6 | Access & egress | Scaffolding tag missing or showing an expired inspection date |
| 7 | Manual handling | Employee twisting while lifting from a low pallet |
| 8 | Manual handling | Team using a mechanical aid for a load flagged in the risk |
| 9 | Plant & vehicles | Pedestrian crossing a forklift route outside the marked walkway |
| 10 | Plant & vehicles | Driver completing a full pre-use check before first movement |
| 11 | Machinery | Interlock guard bypassed or taped over |
| 12 | Electrical | Damaged extension lead or missing PAT label |
| 13 | Working at height | Ladder used at the wrong angle or not footed |
| 14 | Environment | Poor lighting in a stairwell or loading bay |
| 15 | Behaviour | A supervisor stopping work when conditions changed mid-task |
Six of these fifteen are positive observations. That ratio is deliberate – it’s roughly what a mature programme looks like.
Why do safety observations matter?
Injury rates tell you what already happened. Observation data tells you what is about to happen.
The scale of the problem in Great Britain is well documented. HSE reported 1.9 million workers suffering from work-related ill health and 40.1 million working days lost in 2024/25, alongside 59,219 employee injuries reported under RIDDOR. A further 126 workers were killed in work-related accidents in 2025/26. Every one of those sat downstream of conditions and behaviours that were, at some point, observable.
What we see in the platform supports that. Across Notify customers in 2025, users logged 256,784 safety events and completed 767,353 audits and inspections. The pattern in the accounts that improve fastest is consistent – reporting volume goes up first, injury rates come down after.
- Menzies Distribution Solutions – a 40% reduction in lost time accidents and RIDDOR-reportable incidents, alongside a sustained rise in safety observations coming from a much wider range of employees rather than the usual few.
- Stadler Rail UK – an 87% year-on-year increase in close call and near miss reports, and a 37% reduction in lost time incident rate within six months.
There is also a regulatory case. The Management of Health and Safety at Work Regulations 1999 require employers to plan, organise, control, monitor and review their preventive measures. ISO 45001 clause 9.1 requires monitoring and measurement of OH&S performance. Observation data is one of the most straightforward ways to evidence both.
How to submit a safety observation report
A good report gives the person closing it out everything they need. With Notify’s Incident Management Software, it takes just 60 seconds to submit a report.
- Record what you saw, not what you concluded. ‘Guard removed on press 4’ beats ‘unsafe machining practice.’
- Add the location and time. Vague locations are the single biggest cause of observations that can’t be actioned. With Notify, you can geo-tag your exact location.
- Attach a photo. It removes ambiguity and speeds up close-out.
- Say whether it’s an unsafe finding or a positive observation. This is what makes the data analysable later.
- Note any immediate action taken and describe the situation.
- Assign an owner and a due date. Hit submit, and you’ll receive an email confirmation of your report having been submitted.
How to measure a safety observation programme
These are the four metrics worth putting in front of a board. Set your own baseline in the first quarter, then track the trend.
| Metric | How to calculate | Worked example | What it tells you |
| Observation rate | (Observations in the month ÷ headcount) × 100 - expressed as observations per 100 workers per month | 240 observations, 800 staff → 30 per 100 workers | Whether people are engaging at all. Normalising per 100 workers lets you compare a 60-person depot against a 600-person site fairly. |
| Participation rate | (Unique reporters in the month ÷ headcount) × 100 | 96 individuals reported, 800 staff → 12% | Whether reporting is broad or carried by a handful of people. A high observation rate with a low participation rate means you have five enthusiasts, not a programme. |
| Close-out rate | (Actions closed in the period ÷ actions raised in the same period) × 100 | 180 closed of 240 raised → 75% | Whether reporting leads to change - the metric that determines whether reporting keeps happening. |
| Positive ratio | (Positive observations ÷ all observations) × 100 | 72 of 240 → 30% | Whether the programme is seen as recognition or enforcement. |
Close-out rate is the one to watch. When people report something and nothing changes, they stop reporting, which is detrimental to your proactive safety culture.
How does Notify help you capture and act on safety observations?
Notify is a proactive safety management platform used by over 300,000 workers globally.
Report in seconds from the floor. Frontline workers submit observations from the mobile app with photos, geo-tags, and notes – online or offline.
Nothing gets lost. Every observation becomes a tracked action with an owner and a due date through Action Tracking.
See the pattern, not just the report. Safety Intelligence Dashboards show observation rates, close-out performance and emerging trends by site, team and category – ready for your board.
Notify Spark, our AI companion, automatically surfaces insights and themes across your observation data.
Book a demo or try Notify for free.
FAQs
A near miss requires an event that could have caused harm but didn’t. A safety observation doesn’t need an event at all – it records a condition or behaviour someone noticed. Every near miss is worth reporting; not every observation is a near miss.
Everyone. Programmes that restrict observations to managers and safety staff generate a fraction of the data and miss the tasks only the people doing them can see. The most effective programmes make reporting available to every worker, including contractors. With Notify, anyone can report an observation – even those without a Notify login.
There is no universal benchmark, and any figure quoted as one should be treated with caution. Set a baseline over your first quarter, then judge the programme on the trend in your observation rate, participation rate, and close-out rate.
There is no regulation that says “carry out safety observations.” But the Management of Health and Safety at Work Regulations 1999 require employers to monitor and review preventive measures, and observations are a recognised way of evidencing that you do.
A record of something being done safely, for example, correct PPE, a proper pre-use check, or someone stopping a job when conditions change. Positive observations confirm which controls work in practice and make reporting feel like recognition rather than enforcement.