Picture a site trailer at a quarry. There is a monitor bolted to the wall above the coffee maker, and on it a grid of gauges — recordable rate, near-miss count, days since the last lost-time injury, training compliance. Almost all of them are green. It is seven in the morning. That same morning, a loader operator backed into a berm hard enough to rattle the cab, and nothing on that screen will know about it for another three hours. The gauges are green now. They will still be green at lunch.

I have stood in front of a screen like that. I have also helped build them. So when I tell you I am skeptical of EHS dashboards, understand that I am most skeptical of my own first instincts.

The glance is not the job

A dashboard optimizes for the glance. That is the whole assignment: compress a week of operational reality into something a busy person can absorb in four seconds on the way to a meeting. For one kind of question — is anything actively on fire — the glance earns its keep. The trouble starts when the glance becomes the product. When the green light is treated as the deliverable.

Because a safety program does not run on glances. It runs on what a supervisor decides to do differently before the next shift. And a wall of green gauges is exquisitely designed to suggest that the right answer is "nothing."

There is a deeper problem underneath the design one. The numbers most EHS dashboards lead with are lagging by construction. Recordable rate, lost-time frequency, days-since — these are scoreboards for games that already finished. They tell you how last quarter went. They are necessary for the regulator and close to useless for the crew clocking in at six, because by the time one of those numbers moves, the event it measures already happened to a person with a name.

What we built instead

So when we built our own platform — and yes, the fact that what we make is, near enough, a dashboard too is not lost on me — we tried to invert the default. The organizing question we kept returning to was not "what happened" but "what should happen next, and who owns it."

A finding from an inspection does not resolve into a chart. It resolves into a structured analysis with corrective actions — each one tied back to the specific standard it answers to, each one assigned to a person, each one in a state that is either open or closed and cannot be quietly rounded up to green. The screen's job, the way we think about it now, is to put the next action in front of the person who has to take it. Not to reassure the person three levels up that the trend line is behaving.

What I got wrong

I did not start there. The first internal version of the overview screen was, I am a little embarrassed to say, a beautiful wall of charts. Stacked bars by category. A trend line with a tasteful gradient fill. A composite safety score in a big ring that swung from amber to green as you "improved." I was proud of it.

I showed it to an EHS director I trust, expecting her to admire the composite score most of all. She looked at it for a while and asked the question that reorganized the entire project: "Okay. What do I do on Monday?"

I did not have an answer the screen could give her. The score told her she was a 78. It did not tell her that four corrective actions were past due at the south pit, or that the same guard had been written up twice in a month, or which of the things on that board she could actually move before the next shift. The composite, in particular, turned out to be the worst offender. It took a dozen real, specific, fixable conditions and laundered them into a single number that felt like information and functioned like a sedative. We tore it out. I have not built another one since, and I do not intend to.

What the dashboard refuses to do

Which brings me to the parts of the platform that are defined by what they will not do.

It does not roll your program up into a single grade. No one number, no letter, no traffic light for "your site." We decided early that any composite which can read green while a corrective action sits open is a composite that lies — and we would rather show you the open action.

It does not gamify. No streaks, no badges for days-since, nothing that turns the most underreported category in all of safety — the near-miss — into a number people are quietly rewarded for keeping low. The fastest route to a flattering near-miss chart is a crew that has learned not to report, and a screen that celebrates the chart is working against the program it claims to serve.

And it does not show you green while something is open. A closed-out finding is green. An open one stays open, in plain sight, until someone does the work. The screen is not allowed to be more comfortable than the site actually is.

The test that actually matters

None of this means the dashboard problem is solved. I still catch myself reaching for the satisfying chart, the summary number, the view that makes a hard week look tidy. The discipline is in resisting it — in remembering that the measure of one of these screens is not how informed the executive feels, but whether the supervisor at the south pit did one thing differently because of what it showed her. I am more interested in that second test than I have ever been, and I think it is where the next few years of this work quietly live.

If you have had the "what do I do on Monday" moment — the dashboard that looked great and changed nothing — I would like to hear about it. The standing offer holds: thirty minutes against your actual MSHA or OSHA inspection history. You bring last year's findings; we spend the time on the actions and who owns them, not on the gauges. No deck, no demo theatre.

Get in touch

Glad to be writing this one. Next post when something is worth saying.

— Jeffrey