Hazards, incidents, and near-misses
Three different things often get blurred in safety conversations:
- A hazard is a condition that could cause harm.
- An incident is an event that did cause harm.
- A near-miss is an event that could have caused harm but didn’t.
SteadyOn separates them — hazards live in one module, incidents (including near-misses) in another. This page explains why.
Hazards: things that could go wrong
Section titled “Hazards: things that could go wrong”A hazard is a standing condition. It’s there whether or not anyone’s been hurt by it. Examples:
- Wet kitchen floor during wash-down.
- An angle grinder without a guard.
- Working at heights on the second floor.
- Aggressive customer behaviour at the front desk.
Hazards aren’t always physical. Psychosocial hazards (bullying, unreasonable workload, customer aggression) are real hazards.
Hazards live until they’re closed. They get re-reviewed periodically. They’re the single most useful register for an inspector or auditor to read because they show what you know about and are managing.
Incidents: things that did go wrong
Section titled “Incidents: things that did go wrong”An incident is an event with a date and time. Something happened. Someone tripped, an arm was sprained, a chemical spilled, a fire alarm sounded. Each incident is a one-off record.
Some incidents are notifiable — the regulator must be told. The threshold varies by jurisdiction (see HSWA 2015 for NZ specifics).
Near-misses: things that almost went wrong
Section titled “Near-misses: things that almost went wrong”A near-miss is an event where harm could have occurred but didn’t — the worker slipped but caught themselves; the chemical container fell but didn’t break; the chainsaw kicked back but missed.
In SteadyOn, near-misses are recorded as incidents with type Near-miss. They’re not a separate module because:
- They have a date, a time, and a location — they’re events, not conditions.
- They follow the same investigation lifecycle as a real incident.
- The pattern of near-misses is one of the most valuable signals in any safety programme; mixing them in with real incidents makes them visible.
Why split hazards from incidents
Section titled “Why split hazards from incidents”A novice safety system often blurs them: “we had a slip incident, the hazard is wet floors”. They feel related — and they are linked in SteadyOn — but they’re different things:
| Hazard (condition) | Incident (event) |
|---|---|
| Always there | Happened on a date |
| Open until closed | Investigated and closed |
| Re-reviewed periodically | One-off |
| Carries a risk level | Carries a severity rating |
| Has controls in place | Has root cause + actions |
The same wet kitchen floor can be one hazard and ten incidents over a year. Splitting them lets you ask each side a meaningful question:
- “What hazards do we know about, and how are we managing them?”
- “What incidents have we had, and what have we done about each?”
If they’re collapsed into a single record, both questions get muddy answers.
Why link them anyway
Section titled “Why link them anyway”You should link incidents back to hazards via the Related tab on either side. The link is what makes the story cohere.
When an inspector asks “what’s your worst slip incident?” you can open it, click through to the hazard, and show the assessment, the controls, and the review schedule that was already in place. Without the link, you’d answer with two unrelated screenshots.
When an inspector asks “what hazards do you have around chemical handling?” you can open the hazard, click through to its three incidents over the past year, and show how each was investigated and turned into corrective actions.
Near-misses are gold
Section titled “Near-misses are gold”The pyramid in safety folklore — for every fatality, X serious injuries, Y minor injuries, Z near-misses — is debated in detail but right in spirit. There are many near-misses for every actual harm. Reporting them gives you a richer signal than waiting for real injuries.
To make near-miss reporting work, the friction has to be near zero. SteadyOn’s public reporting link is designed for exactly this — anyone can report a near-miss in 30 seconds without a login.