A site with no near-miss reports for six months has one of two things going on. Either the work has become safe, or the people doing it have stopped telling you about it.

Most senior HSEQ leaders will recognise which of the two is more likely. Even so, the quiet site often receives a lighter audit schedule, fewer leadership visits and a mention in the board pack as an example of strong performance. Silence can read as success because, on a dashboard, the two look the same.

The difference becomes more obvious in the investigation after an event. A serious incident happens at a site that had reported nothing, and the interviews reveal a workforce who had seen the hazard for months. They had worked around it, talked about it in the cabin and decided it was not worth raising. By the time the incident report is written, the weak signals may have been visible on site for a year. This is what near miss underreporting looks like from the boardroom.

Why people stop reporting

The reasons rarely involve bad intent. Frontline teams stop reporting when the last three reports produced no visible action. They stop when raising a concern means a long form on a slow system at the end of a shift. They stop when a supervisor under delivery pressure sighs at the mention of a close call. And they stop when the organisation measures them on frequency rates, because every report they file looks like a mark against their own team.

Psychological safety in a high-risk environment comes down to a simple test. Can a rigger, a track worker or a stevedore say “I nearly got hurt doing that” without any cost to themselves? If the answer is no, your reporting data describes your culture rather than your risk.

The sign at the gate that says 412 days without an incident

Many sites display one. A counter at the entrance, a zero-harm target in the annual plan, a dashboard in the site office showing days since the last recordable. Leadership teams put them there to make safety visible, and they do. They also make something else visible to every person walking under them: the number is the point, and you are the one who could reset it.

When that counter is tied to a bonus, a team award or a contractor’s scorecard, the incentive is explicit. A sprained wrist reported on day 412 costs the whole crew money. A near miss that triggers an investigation puts the site’s performance rating at risk and the contractor’s chance of the next tender with it. People respond to incentives the way people always have. The injury gets strapped up in the van, the near miss gets discussed over a brew and never written down, and the counter keeps climbing.

The result is a site that looks safer than it is, measured by an indicator the workforce has every reason to protect. A zero on the board tells you nothing has been reported. It cannot tell you nothing has happened. The two drift further apart with every month the counter runs, because the longer the streak, the more it costs to be the person who ends it.

None of this means zero harm is the wrong ambition. It means the ambition should sit on outcomes nobody can suppress (fatalities, permanent injuries, barrier failures found in audit) and never on counts the workforce controls by staying silent. If you reward anything connected to reporting, reward the reporting.

Cognitive load makes quiet sites quieter

There is a second reason weak signals go unspotted, and it has nothing to do with willingness. People under heavy mental load miss things. A control room operator juggling permits, radio traffic and a shift handover has less attention available for the small deviation that would have told you a barrier was degrading. Conflicting priorities, risks that nobody has explained properly and the general burden of running complex work all reduce the capacity to notice, let alone report.

This is where the gap between work as imagined and work as done becomes a safety issue. The procedure describes a sequence that assumes time, equipment and attention that the shift does not have. People adapt, quietly, every day. Those adaptations are the richest source of information about where your controls are weakest, and almost none of them appear in your incident system.

What leaders can do with this

The organisations that get ahead of this treat silence as a signal in its own right. Three practices make the difference:

• Close the loop visibly. Every report should produce a response the reporter can see, even if the response is a decision to accept the risk. People keep reporting when they can see it changes something.

• Use Learning Teams or Human and Organisational Performance (HOP) conversations to surface how work is done, rather than relying on observation cards to capture deviations from how it was imagined.

• Take incident counts out of rewards. Remove days-without-incident targets from bonuses, team awards and contractor scorecards, and recognise teams for the quality of what they report instead.

Is your reporting data telling you about risk, or about culture?

Before looking at tools, it is worth asking a few questions of your own organisation. The answers tend to show whether a quiet site is quiet because the work is safe or because people have stopped reporting.

• When near miss reports fall at a site, does anyone treat that as a signal, or does it register as good news?

• Can a frontline worker see what happened to the last concern they raised?

• How long does it take to report a near miss from the place where it happened, and does it get harder in poor signal or at the end of a shift?

• Are any bonuses, awards or contractor scorecards linked to incident counts or days without an incident?

• Do your investigations ask what the person was dealing with at the time, or only whether they followed the procedure?

• Could you tell, today, which of your sites has gone quiet in a way its activity does not explain?

If more than two of those are uncomfortable to answer, the quiet sites deserve a closer look.

How COMET helps

Finding the causes behind failures and stopping them recurring is what COMET exists to do, and that work depends on the information frontline teams are willing to share. Three things help.

COMET’s incident management module captures observations and near misses on a mobile device in the time it takes to send a text, with a configurable risk matrix so that minor reports do not trigger major paperwork. When reporting costs nothing, people report. Every report then gets a visible response, which is what keeps the next one coming.

COMET scales the depth of investigation to the actual and potential severity of the event. For minor incidents and near misses, the COMET Lite process provides a streamlined, structured approach that lets teams complete investigations efficiently, so reporting a close call never turns into a paperwork burden. For events with higher actual or potential severity, such as lost time injuries, serious near misses, or significant quality or environmental concerns, the full COMET methodology is applied. This includes a comprehensive Human Factors Analysis, which focuses on understanding the real-world context and pressures faced by those involved, rather than simply checking procedural compliance.

Both COMET Lite and full investigations use the same structured root cause taxonomy, ensuring that systemic issues, such as cognitive load, conflicting priorities, or unclear risk are consistently captured and visible across your data, regardless of the investigation route. This structured approach means that key learnings are not lost in free-text fields, but are available for trend analysis and organisational learning. If a COMET Lite investigation reveals greater severity or complexity than initially assessed, the process can be escalated to a full COMET investigation to ensure appropriate depth and rigour.

COMET Signals analyses your whole history of HSEQ records, structured fields and free-text narratives alike, and shows what a monthly KPI pack tends to miss. Thinning reporting shows up in two ways: a site filing fewer reports than its own recent baseline, and reports shrinking to a single line with no account of why the event happened. Signals scores every report on quality and causal depth, so the second kind of silence is as visible as the first. Recurring themes surface even when nobody uses the same words twice, because every narrative is classified against the same taxonomy. And silence gets context: event trends sit alongside the operations, management-of-change and contractor data you load, so a fall in near misses during a site’s busiest month stands out. Set a tracker on any site’s reporting rate and Signals alerts you when it drops below the threshold you choose.

That gives you the evidence to ask the right question at the right site, before an investigation asks it for you.

If you want to understand what your quiet sites might be telling you, book a discovery call.