All Resources
A mid-sized manufacturer with four plants and around 1,200 employees spent two years trying to fix a safety program that looked fine on paper. Incidents were logged. Audits were scheduled. Reports went to the board every quarter. Yet the same failures kept recurring across sites, and nobody could say why. This case study walks through what was actually broken, how the rollout was sequenced, and what changed in the numbers over eighteen months.
The Situation
The company ran three separate systems. Incidents lived in a legacy database installed in 2009. Audits ran on spreadsheets, one file per site, emailed monthly. Corrective actions were tracked in a project management tool the maintenance team had adopted for something else entirely.
Each system worked. None of them talked. An incident could be closed in the database while its corrective action sat open in another tool, owned by someone who had changed roles. Nobody noticed, because no single view existed.
The EHS manager described the quarterly board report as three days of copy and paste that still could not answer whether the company was getting safer.
What Was Breaking
A two-week review across the four plants found the failures were structural, not personal.
- Near-misses were invisible. The reporting form was a PDF, printed and left in a tray by the shift office. The average plant logged four near-misses a month. For a workforce that size, the real number should have been closer to forty.
- Actions expired silently. Thirty-one percent of corrective actions were past their due date. No escalation existed, so overdue simply meant forgotten.
- Root causes were not shared. A guarding failure fixed at Plant 2 recurred at Plant 4 eleven months later. The lesson never crossed sites.
- Audit evidence took weeks to assemble. Preparing for a customer audit consumed roughly two working weeks of the EHS manager's time.
The Rollout Plan
The team deliberately rejected a single big launch. Rolling out four modules across four plants at once would have meant training everyone on everything, while competing with production for attention.
Instead they sequenced by volume and risk. Start with the workflow that has the highest volume and the lowest consequence of getting it wrong. Prove it works. Then move up.
One plant was chosen as the pilot, specifically the one with the most sceptical supervisors. The logic was straightforward: a rollout that convinces the hardest audience will survive everywhere else.
Phase One: Incident and Hazard Reporting
Phase one covered hazard and near-miss reporting only, on mobile, at the pilot plant. Nothing else changed.
The single biggest decision was removing the login. Workers scanned a QR code posted at each work cell, which opened a pre-identified form. Reporting a hazard took about forty seconds, including a photo.
Near-miss reports at the pilot plant went from four a month to fifty-two in the first full month. The EHS manager's first reaction was that something was wrong with the data. It was not. The hazards had always been there. The reporting friction had been hiding them.
Phase one ran for ten weeks before anything else was added.
Phase Two: Audits and Corrective Actions
Phase two connected the reports to action ownership, then moved audits off spreadsheets.
Every hazard and incident now routed automatically to a named owner with a deadline. Overdue actions escalated to the plant manager after seven days, and to the operations director after fourteen. The escalation changed behaviour, not the tracking.
Audits moved to mobile checklists with photo evidence attached at the point of inspection. Findings generated corrective actions directly, with no rekeying.
Rollout to the remaining three plants followed the same two-phase sequence, one plant per quarter.
What Changed in the Numbers
Eighteen months after the pilot began, measured across all four plants:
- Near-miss reporting rose from 16 to 214 per month across the group. Leading indicator volume was the clearest signal the culture had shifted.
- Overdue corrective actions fell from 31 percent to 6 percent. Escalation did most of this work.
- Average incident closure time fell from 9.2 days to 3.1 days.
- Recordable incidents fell 34 percent year over year.
- Audit pack preparation went from roughly two weeks to under two hours.
- Repeat root causes fell by roughly half, because findings from one plant were now visible to all four.
The recordable reduction was the number the board cared about. The EHS team cared more about near-miss volume, because that was the input that produced it.
What the Team Would Do Differently
Asked what they would change, the team named three things.
- Involve supervisors earlier. Supervisors absorbed the largest workload change and were consulted after the pilot design was fixed. They should have been in the room first.
- Set the reporting expectation explicitly. Some supervisors initially read rising near-miss numbers as their plant getting worse. Leadership had to state clearly that more reports meant better visibility, not worse performance.
- Migrate less history. Three months were spent moving a decade of legacy incident records. In practice, almost nobody queried anything older than two years.
Conclusion
The turnaround did not come from a feature. It came from removing friction at the point of reporting, then making ownership impossible to ignore. Everything else followed from those two changes.
The sequencing mattered as much as the software. Starting with the highest-volume, lowest-risk workflow gave the team a visible win before asking anyone to change how they ran audits.
Neptune EHS supports this pattern directly, with QR-based reporting, automatic routing, escalation rules, and cross-site trend visibility in one system. If you are planning a similar rollout, starting with one plant and one workflow is the approach this team would recommend.
