The hidden cost of safety complacency in mature organisations

I want to talk about one of the most counterintuitive truths in workplace safety: a good safety record can be the most dangerous thing in your organisation.

Not because good performance doesn't matter. It does. But because sustained good performance creates the conditions for complacency. And complacency, in a mature organisation with established processes and experienced teams, is almost invisible until something goes catastrophically wrong.

SOURCES

The BP lesson: when good numbers hide real risk

The defining case study for safety complacency remains the 2005 BP Texas City refinery explosion. Fifteen people died. More than 170 were injured. The property damage alone exceeded $200 million, and total costs including settlements reached $2.1 billion.

But here's the detail that matters most for this conversation: before the explosion, BP Texas City had an excellent personal safety record. Their injury rates were significantly below the industry average. Executives received bonuses based on these low injury rates. By every metric the organisation was tracking, Texas City looked like a safe refinery.

The Baker Panel report, commissioned after the disaster, identified the central failure with devastating clarity: "BP mistakenly interpreted improving personal injury rates as an indication of acceptable process safety performance at its U.S. refineries." The panel found instances of complacency toward serious risks at every one of BP's five U.S. refineries, not just Texas City.

The Baker Panel's language on complacency is worth dwelling on: "The passing of time without a process accident is not necessarily an indication that all is well and may contribute to a dangerous and growing sense of complacency. When people lose an appreciation of how their safety systems were intended to work, safety systems and controls can deteriorate, lessons can be forgotten, and hazards and deviations from safe operating procedures can be accepted."

The contractors who died in the explosion had just returned from a lunch organised to celebrate one month without a lost-time injury.

What the safety plateau looks like from the inside

BP Texas City is an extreme example, but the underlying pattern is everywhere. EHS Today describes it precisely: comprehensive management systems and improved regulations have helped reduce injury frequency, but this success has led management and safety committees to become complacent. Supervision levels get reduced. Leadership spends less time on the floor. Workers know where the hazards are but aren't reporting them systematically. Many hazards are still present, but the effort to uncover them has stalled.

The result is what the industry calls a safety plateau: a period where injury rates stabilise at a level that's acceptable (but not zero), improvement stalls, and the organisation settles into a maintenance mindset rather than a continuous improvement mindset.

From the inside, the plateau feels like success. Your TRIR is steady. Your audit results are satisfactory. Your near-miss reporting looks reasonable. Nobody is raising alarms. The board is satisfied. The insurance premiums are manageable.

But beneath the surface, several things may be happening simultaneously. Near misses are going unreported because workers see no visible response when they do report. Supervision has thinned because the good numbers suggest it's not needed. New workers are being onboarded faster, with less mentoring, because "we haven't had a serious incident in years." And the organisation is measuring the wrong things: lagging indicators that tell you about the past, not leading indicators that tell you about the future.

The RS 2025 Health & Safety Report captures this dynamic in current data: confidence among EHS professionals in their organisation's ability to protect workers has declined, while concerns about complacency and budget cuts have increased. Safety culture scores are dropping. Working days lost to injury remain in the tens of millions.

Why mature organisations are especially vulnerable

Complacency doesn't happen in organisations that know they have a problem. It happens in organisations that believe they've solved the problem.

Mature operations with established safety programmes, experienced teams, and a track record of good results are particularly susceptible for several reasons.

Success breeds assumption. When your injury rate has been stable for three years, it's natural to assume your controls are working. But stable injury rates may simply mean you've harvested the easy improvements (better PPE, basic training, obvious hazard removal) and haven't yet addressed the deeper, systemic risks that cause serious injuries and fatalities.

Metrics can mislead. The ASSP's 2025 trends analysis identifies a significant industry shift away from TRIR toward SIF (serious injury and fatality) and PSIF (potential SIF) metrics. The reason? TRIR can improve while SIF risk remains unchanged or even increases. An organisation can have a declining TRIR while the conditions for a catastrophic event are quietly building. This is exactly what happened at BP Texas City.

Budget follows performance. When safety numbers look good, it's hard to argue for increased investment. The HBR analysis of safety as a performance driver found that most executives frame safety as a compliance issue, see it as a cost, and consequently underinvest. In a mature organisation with a good record, the pressure to reduce safety spending is even stronger because the numbers suggest the current investment is sufficient.

Institutional knowledge erodes silently. Experienced workers retire. Supervisors who remember the last serious incident move on. New team members arrive with no memory of why certain controls exist. Without a mechanism to continuously reinforce the reasons behind safety practices, the practices become rituals performed out of habit rather than understanding. And rituals, once their purpose is forgotten, are easy to shortcut.

The gap between what you measure and what's actually happening

This is where I think the conversation needs to go for organisations that are genuinely serious about getting past the plateau.

The core problem with complacency isn't that people stop caring. It's that the organisation stops seeing. Manual near-miss reporting declines because nobody responds visibly when reports are filed. Safety walks become routine rather than investigative. Audits check boxes rather than challenge assumptions. The data that would reveal emerging risk simply isn't being generated.

When you have no visibility into the near misses, the close calls, and the at-risk behaviours that happen between injuries, you have no way to know whether your safety culture is genuinely healthy or just asleep.

This is exactly the gap that computer vision AI fills. Not as a surveillance tool, but as a visibility tool. When cameras are continuously detecting safety events across your highest-risk areas, complacency becomes measurable. You can see whether near-miss events are increasing even while injury rates remain flat. You can see whether specific zones are deteriorating. You can see whether coaching conversations are happening and whether they're having an effect.

At inviol, we've worked with organisations that came to us precisely because they'd plateaued. Their injury rates were stable, their audits were clean, and their leadership was asking "what else can we do?" The answer, almost always, was: "You need to see what you're currently not seeing."

When we deployed our system and heatmaps revealed hundreds of safety events per week that were completely invisible to their existing reporting system, the plateau made sense. The risk hadn't gone away. It had just gone underground.

Breaking through: from maintenance to continuous improvement

Getting past the plateau requires three shifts that are straightforward to describe and genuinely difficult to implement.

Shift what you measure. Move from lagging indicators (injury rates, lost-time days) to leading indicators (near-miss frequency, safety event trends, coaching completion rates, risk scores by zone). Leading indicators tell you what's about to happen, not what already happened. They're the early warning system that complacent organisations lack.

Shift how you respond to data. The ASSP's 2026 Corporate Listening Tour identifies safety as a value, not a metric, as one of the five critical themes for the future of workplace EHS. That means using data for coaching and learning, not just reporting. When a safety event is detected, the question isn't "who do we blame?" but "what can we learn, and what should we change?"

Shift the conversation with leadership. The organisations that break through the plateau are the ones where safety has a seat at the operations table, not just the compliance table. When you can show leadership that a layout change reduces both safety events and forklift travel time, or that a shift changeover adjustment improves both safety and dock utilisation, safety becomes an operational advantage rather than a cost centre.

The uncomfortable question

Here's the question I'd ask any leader of a mature organisation with a stable safety record: if your current systems are designed to capture less than 10% of the safety events happening on your floor (which is typical for manual reporting), how confident are you that your record reflects reality?

That's not a criticism. It's a genuine question. And the answer, for most organisations, is that they don't know. They can't know, because the data doesn't exist.

The cost of complacency isn't measured in injury rates. It's measured in the risk you can't see, the patterns you can't detect, and the catastrophic event that's building somewhere in the space between what you measure and what's actually happening.

If you're ready to close that gap, book a demo with inviol and we'll show you what your operations look like when you can finally see the full picture. Your numbers might be good. The question is whether they're telling you the truth.

Frequently Asked Questions

What is safety complacency?

Safety complacency occurs when an organisation's sustained good safety performance creates a false sense of security. Teams stop actively looking for hazards, near-miss reporting declines, supervision thins, and the organisation assumes that stable injury rates mean risk is well controlled. The Baker Panel report on the BP Texas City explosion described it as a "dangerous and growing sense of complacency" where "people lose an appreciation of how their safety systems were intended to work."

What is a safety plateau?

A safety plateau is a period where injury rates stabilise and improvement stalls. EHS Today describes it as a state where comprehensive management systems have reduced injuries, but this success leads to complacency, reduced supervision, and a maintenance mindset rather than continuous improvement. The organisation gets stuck in a rut with no visible path to further progress.

How did BP Texas City have good safety numbers before the explosion?

BP Texas City had excellent personal injury rates before the 2005 explosion that killed 15 workers. The Baker Panel found that BP focused on personal safety metrics and mistakenly interpreted low injury rates as evidence of good overall safety. In reality, serious process safety risks were building beneath the surface, undetected by the metrics being tracked.

How does AI help break through a safety plateau?

Computer vision AI provides continuous visibility into safety events that manual reporting misses. By detecting near misses, proximity breaches, and at-risk behaviours automatically, it generates the leading indicator data that complacent organisations lack. This data reveals whether risk is genuinely declining or simply going unreported, and it provides the evidence base for targeted coaching and operational improvements.

What leading indicators should mature organisations track?

The ASSP recommends shifting from TRIR to SIF and PSIF (potential serious injury and fatality) metrics. Leading indicators include near-miss frequency and trend direction, safety event rates by zone and shift, coaching completion and engagement rates, and the ratio of high-severity to low-severity events over time. These indicators tell you what's about to happen, not what already happened.