The Market Already Knows What Your Documents Know
Why operational failures destroy more than their direct cost — and what industrial operators can do about it.
When a major operational failure makes the news, the numbers in the headline — the fine, the settlement, the repair bill — are rarely the whole cost.
A McKinsey study of 498 major operational-risk events at publicly traded North American and European companies (2006–2020, direct losses of at least $50 million) found that in the 120 days after an event became public, affected companies underperformed their peers by an average of 2.7% in total shareholder returns. That represented approximately $1.9 billion in lost market value per event — 3.7 times the average reported direct loss of $500 million.
The study covered operational-risk events across sectors — not process-safety incidents alone — and the multiplier varied sharply by geography, industry, event type, and market conditions. But its most important conclusion has direct relevance for industrial operators: markets do not respond only to the cost of the event. They respond to what the event appears to reveal about the organization behind it.
Investors aren't pricing the incident. They're pricing what the incident reveals.
The share price decline doesn't stop when the direct losses are tallied. It deepens for months. McKinsey's interpretation: investors may assume the reported loss understates the truth, and may read the event as a signal of more general mismanagement — foundational issues that compromise the company's ability to create value.
That suggests a deeper interpretation: the market is not merely pricing the incident. It is pricing the information the incident reveals about the organization. Not "this company had bad luck," but "this company didn't know something it should have known."
Anyone familiar with major incident investigations will recognize the pattern.
The knowledge was often already in the building
Across many major process-safety disasters, investigators find that relevant warnings, scenarios, or lessons had already been documented somewhere: in a PHA finding that was never closed out, a HAZOP scenario from a sister unit, an MOC review that did not propagate, or an earlier incident — sometimes at another company — describing a similar failure mode.
The knowledge existed. It just couldn't get to the decision that needed it.
We call these information barriers. They are a recurring contributor to industrial incidents and unplanned downtime: relevant knowledge exists somewhere in the organization but does not reach the person, system, or decision that needs it. The organization, as a legal and financial entity, "knew." The engineer signing the work permit did not.
This may be part of what investors are pricing when they mark a company down after an operational failure. They're not just punishing the event. They're repricing their confidence that the company's knowledge actually reaches its decisions.
The industry did its part — the knowledge exists
To be clear, this is not a failure of knowledge creation. The process industries have institutions most sectors should envy. The U.S. Chemical Safety Board produces investigation reports of extraordinary depth and candor — public, free, and unflinching about root causes. CCPS and AIChE have spent decades codifying process safety knowledge into guidelines, incident databases, and shared learning programs, and organizations like API and the Mary Kay O'Connor Process Safety Center have done the same. Banking, by comparison, needed shared loss databases like ORX — cataloging tens of thousands of loss events per year — before it could even begin quantifying operational risk. Process safety got there first, and did much of it in the open.
The gap is the last mile. All that shared knowledge — plus a company's own PHAs, HAZOPs, MOCs, incident investigations, and procedures — still lives in tens of thousands of pages of unstructured documents, siloed by facility and readable only by the shrinking population of experts who can hold it all in their heads. When those experts retire, the reasoning behind decades of safeguard decisions walks out the door with them. A CSB lesson from 2007 doesn't fail because nobody wrote it down. It fails because nothing connects it to the work permit being signed at your facility this morning.
The raw material for an industrial hazard-intelligence layer exists — much of it thanks to CSB, CCPS, and their peers. What's missing is the layer that puts it inside every decision.
What it takes to build it
Extracting hazard intelligence from safety documents is not a summarization problem. A risk decision — or a regulator, or a board — needs to know exactly where a claim came from, verbatim, with the evidence chain intact. It needs to know which safeguards were credited against which scenarios, whether those barriers actually exist as described, and which documented findings never made it into the permit-to-work system.
That's the standard we hold ourselves to at uc2c.ai with Hazard Navigator: every extracted finding carries full evidence provenance back to the source document, and every barrier claim is resolvable against a canonical inventory. Trust in the answer has to be engineered in, not asserted.
The takeaway for operators
McKinsey's data carries a clear implication: direct losses are not a complete measure of enterprise exposure. The exact multiplier is not transferable to every company or event. But when markets impose an additional confidence penalty after an operational failure, prevention ROI should be evaluated against total value at risk — not only repair bills, fines, and settlements. The economic value of making existing safety knowledge reachable at the point of decision may therefore be materially larger than a direct-loss-only business case suggests.
That also means this argument belongs in front of your CFO and CEO, not just your process safety team. If you work in process safety and have ever struggled to translate a hazard-study backlog or a knowledge-retention problem into the language of enterprise value, this is that translation. We've condensed it into a one-page executive brief — the McKinsey findings, what they mean for an industrial operator, and the questions a leadership team should be able to answer — written for a finance audience: [Download the executive brief (PDF)]. Send it upstairs.
The precursors to a future incident may already be described — in fragments — across documents your organization owns today. The question is whether those fragments can reach the next decision that depends on them.
Ivo Dujmovic is co-founder and CEO of uc2c.ai, building Hazard Navigator, an AI-native process safety intelligence platform. Source: "Response and resilience in operational-risk events," McKinsey on Risk & Resilience, No. 14, June 2023 (article | full issue PDF).
