A worker steps back from a reversing dumper. The machine stops 40 centimetres from his legs. Everyone breathes out. Someone says "that was close." Nobody writes it down.

Six months later, on the same site, the dumper reverses again. This time there is no 40 centimetres.

The Gap Between "Nearly" and "Not Quite"

Major convictions in Irish and UK construction rarely describe freak accidents. They describe systems that failed incrementally, over months or years, while near misses accumulated in people's memories rather than in any reporting system. The incidents that result in €1.2 million fines and multiple fatalities are almost never the first time something went wrong. They are the first time something went wrong and someone died.

Heinrich's Triangle, first published in 1931 and still used in safety training worldwide, argued that for every major injury there are 29 minor injuries and 300 no-injury accidents sitting beneath it. The specific numbers have been contested since. The underlying principle has not. Serious incidents have precursors. Those precursors happen in plain sight. On Irish construction sites, they tend to get filed under "lucky escape" and forgotten by lunchtime.

The HSA has been consistent on this point for years. Voluntary near-miss reporting is one of the indicators they look at when assessing a company's safety culture. Not because they want paperwork. Because a site that reports nothing is a site where workers have learned that reporting leads nowhere.

What a Near Miss Actually Is

The definition matters because sites routinely misclassify events. A near miss is any unplanned event that did not result in injury or damage but had the potential to do so under slightly different circumstances. The key word is potential. Not probable. Not likely. Possible.

That covers:

  • A load shifting on a telehandler but not falling
  • A worker stepping into a trench edge that crumbles but catches themselves
  • A scaffold fitting that loosens but does not drop
  • An unguarded machine that a worker bypasses without incident

None of these produced an injury. All of them contain the exact mechanism that produces injuries. The machinery involved, the sequence of events, the exposure, the gap in controls. Everything except the outcome.

The outcome is decided by physics and timing, not by how careful anyone was that day.

Why Sites Stop Reporting

Two things kill near-miss reporting cultures. Blame and futility.

Blame is the more obvious one. If reporting a near miss means getting questioned about why you were in that position, why you didn't follow the procedure, whether you had proper training, workers stop reporting. The incident gets absorbed into site folklore instead. "Remember when Declan nearly lost his hand on the block cutter?" That story circulates for years. The block cutter never gets guarded.

Futility is quieter and more corrosive. Workers report. Nothing changes. The form goes into a folder. The folder goes into a drawer. The next site manager clears the drawer. After two or three reports disappear into that process, workers rationally conclude that the form is for the company's benefit and not theirs. Reporting stops.

Both failure modes show up consistently in post-incident investigations. When the HSA or a court examines the run-up to a serious incident, they regularly find workers who knew about the hazard, flagged it informally, and were not heard. Sometimes they find written records of concerns that went unanswered.

What Good Reporting Actually Prevents

The prevention value of near-miss reporting is not theoretical. It is mechanical.

When a near miss is properly documented and investigated, it forces an answer to a specific question: what would have to change for this event to not happen again? That question, answered honestly, produces controls. Guards fitted. Exclusion zones established. Procedures changed. Plant segregated from pedestrian routes.

Those controls are in place before the next event. When the next event happens, the outcome is different. The worker walks away. Nothing gets written down because nothing happened. But something did happen. The control worked.

The alternative is to wait for the event that generates paperwork automatically, because someone is in hospital or dead. At that point the investigation happens regardless, the controls get implemented regardless, and the site also faces prosecution, a fine, civil claims, reputational damage, and the particular weight of knowing the outcome was preventable.

Sites that run effective near-miss reporting systems do not have fewer incidents because they are lucky. They have fewer incidents because they are spending the currency of near misses rather than the currency of injuries.

The Machinery Problem Specifically

Construction sites have a machinery near-miss problem that is distinct from other hazard types. The hazards are often intermittent, the controls are often bypassed for speed, and the consequences when something goes wrong are disproportionately severe.

A scaffolding issue tends to be visible. A guard removed from a concrete saw is not visible from a distance. A reversing vehicle blind spot is invisible by definition. Workers and managers often have no real sense of how close the site has come to a serious incident because the near miss was not observable.

This means machinery near misses have to be actively solicited. Plant operators need to be asked, directly and regularly, whether anything unusual happened during their shift. Near misses involving plant are not the kind of thing people volunteer at the end of a long day. They are the kind of thing that comes out in conversation when someone creates the space for it.

Toolbox talks are the right format. Five minutes, specific equipment, direct question: did anything happen this week that you want us to know about? Not "any safety concerns?" That question is too broad. It invites a shrug. Ask about the specific machine, the specific route, the specific task. You will get answers.

The €1.2 Million Is the Floor

The fines attached to construction fatalities in Ireland and the UK have been climbing. €1.2 million represents a serious conviction, but the financial exposure does not stop there. Civil litigation runs alongside criminal prosecution. Insurance premiums reflect conviction history. Senior managers can face personal liability. Projects get delayed during investigation. Some companies do not survive the combination.

That entire exposure sits downstream of a near miss that was not reported, not investigated, and not fixed.

The near miss cost nothing to fix. The fix would have been a physical control, a procedure change, a conversation, a piece of equipment. Measurable in hundreds or thousands of euro. Against a potential liability measured in millions, plus a worker who does not go home.

Sites that treat near-miss reporting as a bureaucratic obligation generate forms. Sites that treat it as an intelligence system generate safety. The distinction is not philosophical. It shows up in the prosecution records.

Report the 40 centimetres. That is what it is there for.