Reducing vehicle-on-plant incidents by 61%: the data behind the result

Across inviol's customer base, vehicle-on-plant incidents reduce by an average of 61%. That's not a projection or a target. It's a measured result across real facilities, running real operations, with real forklifts and real pedestrians.

When we share that number with prospects, the first response is usually some variation of "that sounds great, but how?" Which is the right question. A percentage on its own is meaningless without understanding what's being measured, why the reduction happens, and whether the result is sustainable. So here's the data behind the 61%.

The scale of the problem

Vehicle-on-plant incidents are among the most consequential safety events in any warehouse, distribution centre, or manufacturing facility. The National Safety Council reports that forklifts were the source of 84 work-related deaths in the US in 2024 and over 25,000 non-fatal injuries in 2023-2024. OSHA estimates that 36% of forklift fatalities involve pedestrians, and that roughly 20% of all forklift accidents involve a pedestrian being struck.

In Australia, Safe Work Australia data consistently identifies vehicle incidents as the leading cause of workplace fatalities, accounting for approximately 42% of all work-related deaths. In the UK, one in five workplace fatalities involves a forklift or industrial vehicle.

These are the events that generate the highest-severity outcomes: crush injuries, amputations, fatalities. And they happen in facilities that have training programmes, signage, designated walkways, and safety procedures already in place. The problem isn't a lack of rules. It's a lack of visibility into what actually happens on the floor between vehicles and people, 24 hours a day, across every zone and every shift.

What we actually measure

The 61% figure is based on the reduction in detected vehicle-on-plant safety events across monitored zones, comparing the baseline period (the first weeks after deployment, before coaching and operational changes take effect) against the performance after interventions are in place.

The events we detect and count include pedestrian-vehicle proximity events (a person and a vehicle closer than the defined safe distance), exclusion zone breaches (a vehicle entering a zone where it shouldn't be, or a pedestrian entering a vehicle-only zone), vehicle speed violations (a forklift or other mobile plant exceeding the defined speed threshold for a zone), and near misses that don't result in contact but represent a failure of the expected separation between people and machines.

Every event is captured by computer vision AI processing video from connected cameras, tagged with a timestamp, location, event type, and a short video clip. This means the baseline and the improvement are measured using the same objective, continuous detection method, not relying on voluntary reporting that might change over time.

This is an important methodological point. When you measure improvement using voluntary incident reports, a declining number could mean either "fewer incidents are happening" or "fewer incidents are being reported." When you measure using continuous AI detection, a declining number means fewer events are occurring, because the detection method doesn't change with reporting culture.

The three mechanisms that drive the reduction

The 61% reduction isn't produced by a single intervention. It's the compound result of three mechanisms working together.

Near-miss visibility that triggers targeted action. Before computer vision AI is deployed, most vehicle-on-plant near misses are invisible to the safety programme. They happen, nobody reports them, and the conditions that caused them persist. Once the platform is live, those near misses become visible, quantified, and localised. The heatmap shows exactly which intersections, dock areas, and traffic zones generate the highest density of vehicle-pedestrian interactions. That specificity drives targeted action: a barrier here, a traffic flow change there, an adjusted delivery window at a specific time. Each intervention addresses a specific, data-identified risk rather than applying a generic safety reminder across the whole site.

The data from our customers consistently shows that the majority of vehicle-on-plant events cluster in a small number of zones. Typically, the top three hotspot zones account for 60-70% of all detected events. This concentration means that targeted interventions in a few key areas produce disproportionate results across the site-wide metrics.

Coaching that changes behaviour in the zones that matter. The second mechanism is coaching. When a supervisor receives a face-blurred clip of a forklift-pedestrian near miss that happened in their team's zone yesterday, the coaching conversation becomes specific, timely, and relevant. "Here's what happened at the Bay 4 intersection during yesterday's afternoon shift. What do we think caused it? How do we make sure it doesn't happen again?"

That conversation is categorically different from a generic toolbox talk about forklift safety. It's grounded in a real event, in a real location, involving the team's own work area. The specificity is what drives behaviour change, because the team can see the connection between the event and their daily actions.

Across inviol's customer base, sites with consistent coaching activity (at least two coaching sessions per week using platform-generated clips) show significantly faster event density decline than sites where coaching is sporadic. The coaching isn't just a cultural nice-to-have. It's a measurable driver of the 61% result.

Operational redesign informed by data. The third mechanism addresses the systemic causes that coaching alone can't fix. When the heatmap shows a persistent hotspot at an intersection, and the data reveals that the events concentrate during a specific time window (morning delivery arrivals, shift changeovers, peak picking periods), the response isn't more coaching. It's an operational change: redesigning the traffic flow, moving a barrier, widening a pedestrian walkway, adjusting the delivery schedule, or separating vehicle and pedestrian traffic in the problematic zone.

These operational changes are often the single biggest contributors to the reduction. A traffic flow redesign at a high-density intersection can eliminate 40-50% of the events in that zone in the first month. Combined with coaching in adjacent zones and sustained monitoring to ensure the improvement holds, these operational interventions produce the compounding effect that drives the portfolio-wide 61% average.

The timeline of improvement

The 61% isn't achieved overnight. The improvement follows a fairly consistent trajectory across our customer base.

Weeks 1-2: baseline establishment. The platform goes live and begins detecting events across connected cameras. The first two weeks establish the baseline: how many events per zone per day, where they concentrate, and what types dominate. This is the "we didn't know that was happening" phase that we describe in our first 30 days overview.

Weeks 3-6: quick wins and coaching launch. Safety teams identify the obvious hotspots from the heatmap and begin making targeted interventions (barriers, signage, traffic flow adjustments). Supervisors start running coaching sessions using platform-generated clips. Event density typically begins declining within the first month, with 15-25% reductions common in the initial six weeks.

Months 2-4: compounding improvement. Coaching becomes habitual. Operational changes are refined based on before-and-after heatmap comparisons. The team starts addressing the second-tier hotspots once the primary ones are resolved. The improvement curve steepens as multiple mechanisms compound.

Months 4-12: sustained performance. The rate of improvement gradually flattens as the major risk sources are addressed. By this stage, the cumulative reduction across vehicle-on-plant events typically reaches the 50-65% range. The focus shifts from rapid improvement to sustaining the gains and catching any new risk patterns that emerge from operational changes (new product lines, seasonal peaks, facility modifications).

Why the improvement compounds

One of the things that makes vehicle-on-plant risk reduction different from other safety improvements is the compounding effect. Each intervention doesn't just reduce events in isolation. It changes the conditions that create risk across adjacent areas.

When you redesign a traffic flow at one intersection, the vehicles that used to create near misses there now travel a safer route, which reduces their interaction with pedestrians across the entire corridor. When you adjust a delivery window to avoid the pedestrian peak, the dock area is less congested for the next two hours, reducing risk in zones downstream of the dock. When a coaching conversation helps a team understand why the exclusion zone at Bay 7 matters, the same understanding carries over to their behaviour in other zones.

This compounding is why the 61% average exceeds what you'd predict from adding up the individual interventions. The whole genuinely is greater than the sum of the parts, because each improvement changes the system, not just the specific location where it was targeted.

What the 61% doesn't include

It's worth being transparent about what the 61% figure covers and what it doesn't.

The 61% measures the reduction in detected vehicle-on-plant safety events across monitored zones. It doesn't capture events in areas where cameras aren't connected (though most customers expand coverage over time as the value becomes clear). It measures events detected by the AI, which means it's dependent on camera angles, lighting, and the specific detections configured for each site. And it measures events, not injuries directly, because the point of leading indicators is to capture the precursor events that predict injuries, not to wait for injuries to occur.

That said, the correlation between leading indicator reduction and injury reduction is strong. Customers who achieve significant event density reductions consistently report corresponding reductions in recordable injuries, workers' compensation claims, and equipment damage costs. The overall inviol customer average of 67% risk reduction and 42% incident reduction over three years reinforces this relationship.

What this means for your operation

If your facility has forklifts, reach trucks, tow tractors, or any other mobile plant operating in proximity to pedestrians, vehicle-on-plant events are almost certainly your highest-consequence risk category. The data from across our customer base says that most of those events are preventable, and the pathway to prevention follows a repeatable pattern: make the near misses visible, identify where they concentrate, coach on the behaviours that contribute, redesign the operations that create the systemic conditions, and measure whether the interventions work.

The 61% is an average. Some sites achieve more, some less, depending on their starting baseline, the complexity of their layout, and how aggressively they act on the data. But the direction is consistent, and the methodology is proven across hundreds of customer sites.

If you'd like to understand what the data would look like for your facility, book a demo and we'll walk you through it.

Frequently Asked Questions

How is the 61% reduction in vehicle-on-plant incidents measured?

The 61% is measured by comparing the baseline event density (detected vehicle-on-plant safety events per zone in the initial weeks after deployment) against the event density after coaching and operational interventions take effect. Events are detected continuously by computer vision AI, which means the measurement is objective and not influenced by changes in voluntary reporting culture. The same detection method is used for both the baseline and the improvement period.

How long does it take to achieve a 61% reduction?

The improvement follows a consistent trajectory. Quick wins and initial coaching typically produce 15-25% reductions within the first six weeks. Compounding improvement through sustained coaching and operational redesign deepens the reduction over months 2-4. Most sites reach the 50-65% range within 4-12 months, depending on their starting baseline and how aggressively they act on the data.

What types of events are included in the vehicle-on-plant metric?

The metric includes pedestrian-vehicle proximity events (people and vehicles closer than the defined safe distance), exclusion zone breaches (vehicles or pedestrians entering restricted zones), vehicle speed violations (mobile plant exceeding zone-specific speed thresholds), and near misses where contact nearly occurred. Each event is tagged with a timestamp, location, and video clip.

What drives the reduction: technology, coaching, or operational changes?

All three, working together. Near-miss visibility from computer vision AI identifies where risk concentrates. Coaching using face-blurred video clips changes behaviour in the zones that matter most. Operational redesign (traffic flow changes, barrier placement, schedule adjustments) addresses the systemic conditions that coaching alone can't fix. The 61% is the compound result of all three mechanisms.

Does reducing vehicle-on-plant events also reduce injuries?

Yes. While the 61% measures leading indicator events (near misses and unsafe interactions), customers who achieve significant event density reductions consistently report corresponding reductions in recordable injuries, workers' compensation claims, and equipment damage costs. Across inviol's customer base, the average overall risk reduction is 67% and the average incident reduction is 42% over three years.