Why safety incentive programs often backfire (and what to do instead)

There's a version of this that I've seen at dozens of sites. The safety manager announces a new incentive programme. If the team goes 90 days without a recordable injury, everyone gets a reward: a gift card, a team lunch, a bonus, a day off. The intention is good. The logic seems sound. Reward safe behaviour, and you'll get more of it.

Three months later, the team hits the target. The gift cards go out. Everyone celebrates. And the safety manager quietly wonders why near-miss reporting dropped by 40% during the same period.

This isn't a hypothetical. It's one of the most well-documented problems in workplace safety management, and it's been a point of debate in the EHS community for decades. Safety incentive programmes that reward outcomes (low injury rates, zero-incident periods) have a persistent tendency to drive underreporting rather than safer behaviour. The result is a workplace that looks safer on paper while the actual risk stays the same or gets worse.

The logic that seems right but isn't

The appeal of outcome-based incentives is obvious. If you want fewer injuries, reward people for having fewer injuries. It's the same logic that drives sales commissions and production bonuses: tie the reward to the result, and people will work harder to achieve it.

The problem is that safety isn't like sales. In sales, the metric you're rewarding (revenue) is the thing you actually want. In safety, the metric you're rewarding (low injury rates) isn't the thing you want. What you actually want is safer behaviour, better hazard identification, and a culture where people report problems before they become injuries. Low injury rates might reflect those things, or they might reflect something else entirely: that people have learned to hide their injuries.

This distinction is the difference between lagging indicators (which measure outcomes that have already happened) and leading indicators (which measure the activities that prevent outcomes from happening). When you reward lagging indicators, you're rewarding the absence of bad news. When you reward leading indicators, you're rewarding the presence of good practice.

The three ways incentive programmes backfire

The research on this is extensive, and the mechanisms are consistent.

Underreporting of injuries. This is the primary failure mode. When a reward is tied to a low injury rate, every reported injury threatens the reward, not just for the individual but often for the entire team. The pressure to keep the record clean creates a powerful incentive to avoid reporting. The safety industry even has a name for this: "bloody pocket syndrome," where workers literally hide injuries to protect the team's incentive. A 2008 report from the US House Education and Labor Committee found that a significant majority of workplace injuries may escape counting in federal surveys, and incentive programmes were identified as a contributing factor.

Research published in Professional Safety (the journal of the American Society of Safety Professionals) documented that outcome-based incentive programmes frequently result in underreporting of injuries and near misses, with workers pressuring each other not to report events, particularly when the reward period is close to completion.

Suppression of near-miss reporting. This is arguably more dangerous than injury underreporting, because near misses are the leading indicators that give you early warning of systemic risk. If workers learn that reporting a near miss could jeopardise the team's incentive, near-miss reporting declines. The irony is that the incentive programme, designed to improve safety, actively destroys the data you need to identify and fix hazards before someone gets hurt.

Distraction from root causes. Outcome-based incentives focus attention on the metric (the injury rate) rather than on the conditions that produce injuries. If the reward is for zero incidents, the organisational energy goes into maintaining the number, not into investigating why near misses happen, redesigning traffic flows, or addressing the layout problems that create risk. As one critical assessment published by the International Risk Management Institute noted, incentive programmes that don't require any change in processes or procedures are essentially rewards for maintaining the status quo.

What OSHA says (and changed its mind about)

The regulatory position on safety incentives has shifted over time, reflecting the complexity of the issue.

In 2016, OSHA's Final Rule to Improve Tracking of Workplace Injuries and Illnesses effectively restricted outcome-based incentive programmes, stating that they could discourage reporting of injuries and illnesses without improving safety. In 2018, OSHA reversed course with a memorandum clarifying that the agency does not prohibit rate-based incentive programmes, provided the employer has implemented adequate precautions to ensure employees feel free to report injuries.

The practical takeaway: OSHA allows outcome-based incentives but explicitly acknowledges the risk of underreporting and requires employers to demonstrate that reporting is not being suppressed. For organisations operating under New Zealand's HSWA or Australia's WHS framework, the principle is similar: regulators want evidence of genuine risk management, not artificially low numbers.

What to do instead: reward the inputs, not the outputs

The alternative isn't to abandon recognition entirely. Recognition matters. People who feel their safety contributions are valued are more engaged, more vigilant, and more likely to participate actively in the safety programme. The shift is from rewarding outcomes (low injury rates) to rewarding the activities that produce safe outcomes.

Reward reporting, not silence. Instead of rewarding zero incidents, recognise workers who report near misses, identify hazards, and raise safety concerns. A team that submits 20 near-miss reports in a month is demonstrating exactly the vigilance you want. That deserves recognition, not a penalty for having "too many" reports. With computer vision AI detecting events continuously, you have an objective baseline of how many events are occurring. A voluntary reporting rate that tracks close to the AI-detected event rate tells you your culture is healthy.

Recognise coaching participation. Supervisors who consistently run coaching conversations using real event data, and teams that actively engage in those conversations, are doing the work that prevents injuries. Recognise the coaches and the participants. Track coaching session completion rates as a leading indicator and make it visible. When people see that coaching is valued, coaching happens more consistently.

Celebrate hazard elimination. When someone identifies a layout problem, proposes a traffic flow change, or flags a blind corner that the heatmap confirms is a hotspot, that's a safety contribution worth celebrating. It's specific, constructive, and directly connected to risk reduction. Recognise the individuals or teams who identified the problem and the ones who implemented the fix.

Use team-based recognition for leading indicator improvement. Rather than rewarding the absence of injuries, reward the improvement in measurable leading indicators. If Zone C's event density dropped 30% this month after the team implemented a traffic flow change and ran three coaching sessions, that's a concrete, measurable achievement. It's also one that can't be gamed through underreporting, because the detection is continuous and objective.

Measuring what matters

The reason outcome-based incentives persist, despite their well-documented problems, is that many organisations don't have anything better to measure. If your only safety metrics are LTIFR and TRIFR, then those are the only numbers you can tie rewards to.

Computer vision AI changes this equation fundamentally. When you have continuous, objective, facility-wide data on near misses, exclusion zone breaches, speed violations, and pedestrian-vehicle interactions, you have a rich set of leading indicators that can form the basis of a recognition programme built on inputs rather than outputs.

Event density trends by zone, coaching session frequency and completion, time from hazard identification to corrective action, intervention effectiveness (did the change actually reduce events?): these are all measurable, meaningful, and directly connected to the activities that prevent injuries. They're also much harder to game than injury rates, because the data is generated by the AI, not by voluntary reporting.

This is the measurement infrastructure that makes a fundamentally different kind of recognition programme possible. Instead of "congratulations on zero injuries this quarter (we hope nobody's hiding anything)," you can say "congratulations on a 35% reduction in pedestrian-vehicle events in your zone this quarter, driven by the traffic flow change your team proposed and the coaching sessions your supervisors ran." That's a recognition statement grounded in specific, verified, meaningful achievement.

The culture you're actually building

The deepest problem with outcome-based incentives isn't the underreporting (though that's serious enough). It's the culture they create.

When you reward the absence of bad news, you build a culture where bad news is hidden. Where reporting feels like a risk rather than a contribution. Where the safest career move is to stay quiet and hope for the best. That's the opposite of the just culture, the reporting culture, and the learning culture that every serious safety framework (from aviation's SMS to the HSWA's PCBU obligations) is trying to build.

When you reward participation, reporting, coaching, and measurable risk reduction, you build a culture where speaking up is valued, where data is trusted, where supervisors are coaches rather than enforcers, and where every team member feels like an active participant in keeping the operation safe.

That's the culture that produces sustained, year-on-year improvement. Not because people are chasing a gift card, but because they understand the risk, they have the data to see it, and they know their contributions are recognised and valued.

If you'd like to see how continuous safety data supports a recognition programme built on leading indicators, book a demo and we'll walk you through it.

Frequently Asked Questions

Why do safety incentive programmes cause underreporting?

When rewards are tied to low injury rates or zero-incident periods, every reported injury threatens the reward for the individual and often for the entire team. This creates pressure to avoid reporting injuries and near misses. Research consistently documents this effect, including a 2008 US House Education and Labor Committee report and studies published in Professional Safety. The phenomenon is widely known in the safety industry as "bloody pocket syndrome."

Does OSHA allow safety incentive programmes?

Yes. OSHA's 2018 memorandum clarified that rate-based incentive programmes are not prohibited, provided the employer has implemented adequate precautions to ensure employees feel free to report injuries and illnesses. However, OSHA's 2016 Final Rule explicitly acknowledged the risk that such programmes can discourage reporting without improving actual safety. The practical implication is that outcome-based incentives carry regulatory risk if reporting suppression cannot be demonstrably prevented.

What should I reward instead of low injury rates?

Reward the activities that produce safe outcomes rather than the absence of bad outcomes. This includes near-miss and hazard reporting (rewarding vigilance, not silence), coaching participation (recognising supervisors who run regular coaching conversations and teams that engage), hazard elimination (celebrating specific risk-reduction contributions), and measurable improvement in leading indicators like event density trends in specific zones.

How do leading indicators prevent the gaming problem?

Leading indicators generated by computer vision AI (such as event density, exclusion zone breaches, and speed violations) are detected objectively and continuously by the system, not by voluntary reporting. This means they can't be gamed through underreporting. A declining event density trend in a monitored zone reflects a genuine reduction in risk, not a suppression of data. This makes leading indicators a far more reliable basis for recognition programmes than lagging indicators like LTIFR.

Can safety recognition programmes really improve safety culture?

Yes, when designed correctly. Recognition that rewards reporting, coaching participation, hazard identification, and measurable risk reduction builds a culture where speaking up is valued and safety is a shared responsibility. This is consistent with the just culture principles that underpin aviation safety management systems and the proactive risk management expectations of regulators like WorkSafe NZ and Safe Work Australia. The key is rewarding the inputs (behaviours and activities) rather than the outputs (injury statistics).