All Resources
We sat down with three EHS practitioners with a combined fifty years across manufacturing, energy, and logistics. The conversation covered what has genuinely changed in the profession, why well-funded programs still stall, and what they would tell a team starting out today. Their answers have been edited for length.
Meet the Panel
Dana R. runs EHS for a multi-site food manufacturer, fifteen years in the field, previously a plant safety coordinator.
Marcus O. is a former offshore HSE advisor now consulting for energy operators across three continents, twenty-two years in.
Priya S. leads safety and compliance for a logistics group operating 40 depots, thirteen years in, with a background in operations rather than safety.
On the Biggest Shift in the Last Five Years
Marcus: The reporting burden moved off the safety team and onto the system. That sounds small. It is not. I used to spend a third of my week assembling evidence someone else had already collected. That third came back.
Priya: For me it is that the frontline finally has a way in. When I started, reporting a hazard meant finding a supervisor, who found a form. Most people just did not bother. Now it is a phone in their pocket. The volume of what we see went up by an order of magnitude, and none of that was new risk. It was always there.
Dana: Expectations changed. Customers audit us now in ways they did not before. They want to see the trail, not the summary. That pushed a lot of companies to digitise faster than they planned to.
On Why Programs Stall
Dana: Almost always because it was rolled out as a compliance project. If the message is that we need this for the audit, nobody on the floor has a reason to care. The programs that work are framed around the worker, not the auditor.
Priya: Mine is slightly different. Programs stall when nobody owns the actions. You can have beautiful reporting and still go nowhere, because a hazard gets logged, sits in a queue, and expires. Reporting without ownership just creates a very well-documented list of things you did not fix.
Marcus: I would add scope. People try to launch everything at once. Incidents, audits, permits, contractors, all in one quarter. It collapses under its own weight. Pick one thing, get it working, then move.
On Getting Frontline Buy-In
Priya: Close the loop visibly. The first time somebody reports a hazard and sees it fixed within a week, with their name on it, you have them. The second time nothing happens, you have lost them for good.
Dana: We put a screen in the break room showing what was reported and what got fixed. Low tech. It did more for participation than any training session.
Marcus: And do not punish the number going up. The most damaging thing a leader can do is look at rising near-miss reports and ask what went wrong. That single reaction can undo a year of work.
On Measuring What Matters
Marcus: Lagging indicators tell you about last year. They are necessary and almost useless for decisions. If your recordable rate is your main metric, you are steering by the rear window.
Dana: We watch three leading numbers. Reports per hundred workers, percentage of actions closed on time, and how many findings repeat across sites. That third one is the honest measure of whether you are actually learning.
Priya: Time to closure, for me. It is the clearest proxy for whether the system is working. If that number creeps up, something is stuck, and it is usually ownership.
On Where AI Actually Helps
Dana: Pattern spotting. I cannot read 214 near-miss reports a month and notice that eleven of them mention the same walkway. Software can, and it flags it before someone falls.
Marcus: I will be blunt. A lot of what gets sold as AI is a dashboard with a new label. What has real value is triage, sorting signal from routine, and surfacing the thing you would have missed. What does not have value is generating text nobody reads.
Priya: Agreed, with one addition. It helps most where data volume already exceeds human attention. If you are logging four near-misses a month, you do not have an AI problem. You have a reporting problem. Fix that first.
Advice for Teams Starting Out
Priya: Start with one workflow and one site. Prove it. The temptation to go wide immediately is strong and it is almost always wrong.
Dana: Talk to the people who will use it before you choose it. Not after. I have seen platforms bought entirely on a demo that the floor never accepted.
Marcus: Decide what problem you are solving before you look at a single product. Write it down. Then measure against that, not against the feature list you get sent.
Conclusion
Three practitioners, three industries, and a striking amount of agreement. Start narrow. Make reporting effortless. Give every finding an owner. Close the loop where people can see it. And treat rising report volume as the system working, not failing.
Neptune EHS was built around these principles, with mobile-first capture, automatic routing and escalation, and trend detection across sites. If the patterns in this conversation sound familiar, that is usually the sign a program is ready to change.
