Ask your CFO how much cash the company has on hand, and you'll get an answer to the decimal, usually within minutes. Now ask what unplanned equipment failure is likely to cost the business this year: the revenue at risk, the margin exposure, the customer commitments hanging on a handful of critical assets. The answer gets a lot fuzzier. You'll probably get an estimate, maybe a range, and more often than anyone would like to admit, a shrug.

For most of industrial history, that gap was completely excusable, because the data to answer the second question simply didn't exist. But it exists now, and I think that changes what the question actually means. Once an exposure becomes measurable, choosing not to measure it stops being a blind spot and starts being a decision.

Downtime doesn't stay where you budget it

Most enterprises are genuinely disciplined about planned maintenance. It's budgeted, scheduled, tracked against benchmarks, and reviewed quarterly, and finance sees it clearly because it behaves the way a cost is supposed to behave: predictable, categorized, attributable.

Unplanned downtime, on the other hand, doesn't really behave like a cost at all. It behaves more like a leak, seeping into places you'd never think to look for it.

When a critical line goes down, the maintenance portion (the parts, the labor, the emergency contractor) lands in the maintenance budget, right where you'd expect it. Everything else scatters. The overtime to catch up on production shows up in labor. The expedited freight to keep a customer whole shows up in logistics. The output you didn't produce shows up as revenue you never booked, which is to say it shows up nowhere at all. And the customer whose order slipped may quietly re-weight next year's volume toward a competitor, a loss that surfaces eighteen months later as "market dynamics" in a QBR.

Here's the frustrating part: none of those line items actually say "downtime" on them. The maintenance cost of a failure is usually the smallest part of its total business cost, and yet it's the only part that ever gets attributed to the failure.

Why the books can't see it

I want to be clear that none of this is a flaw in your finance team. It's a limitation of what financial reporting was built to do in the first place, because accounting records what happened, organized by category and period. Unplanned downtime scatters its impact across categories and across periods, so no single report ever adds it up.

There's also a second-order effect that's even harder to see: over time, operational instability gets priced into how the business runs. Buffer inventory gets carried because the line might go down. Lead times get padded because delivery can't be promised. Quotes get made conservative because capacity can't be trusted, and capital gets tied up in redundancy that exists only because failure is unpredictable. These are real, recurring costs of unreliability, and they never appear on any report because they've been absorbed into the operating model as "just how we do things."

That's how the true cost compounds while staying invisible. You never feel it as a single event. Instead, you feel it as a business that runs a little heavier than it should, everywhere, all the time.

Industrial AI removed the excuse

For decades, "when will this asset fail?" was an unanswerable question. Machines failed when they failed, and the best a plant could do was inspect on a schedule, replace on a calendar, and keep spares on the shelf. Under those conditions, treating failure as random chance was entirely rational, since you can't put a probability on something you can't observe.

That's no longer the world we operate in. Sensors on critical assets now stream vibration, temperature, and energy data continuously, and AI models trained on failure patterns across thousands of machines read that data and flag developing faults weeks or months before they become breakdowns. None of this replaces the people who keep plants running. Maintenance teams are the unsung heroes of this economy, and the technology's job is to give them superpowers: enough warning to act on their own terms, before a fault becomes a failure. That's what moving from reactive to predictive actually means. Failure risk stopped being an unknowable exposure and became a forecast, with a probability, a timeline, and a cost attached.

And let me be direct here, because "AI" is doing a lot of heavy lifting in a lot of pitches right now. This isn't a model guessing from historical averages. It's physical evidence, a bearing signature changing or a motor drawing more current than it should, detected early enough to actually act on. It's the same shift credit risk went through when scoring arrived. Underwriters didn't disappear; they got better information. The risk itself didn't change, but it became measurable, and once it was measurable, managing it became a discipline instead of a hope.

Which is why I keep coming back to the same conclusion: the blind spot is no longer a data limitation. It's a choice.

What changes when reliability data reaches the CFO

When AI-driven asset health data connects directly to financial performance, reliability stops being a plant metric and starts being a management instrument. In my view, four things change.

Forecast confidence. Revenue commitments rest on production commitments, and production commitments rest on machines. When asset risk is quantified, the company can commit to customers with known confidence instead of hoped-for confidence.

Margin protection. A fault caught early is a planned repair at a planned cost. That same fault caught late becomes emergency labor, expedited parts, scrapped product, and missed shipments, and the difference between those two numbers is margin, often several multiples of the repair itself.

Capital allocation. When you know the actual condition of your assets, you can replace equipment based on evidence instead of age. Capex goes where the risk is, not where the calendar says it should.

Risk posture. Quantified operational risk can be managed the way financial risk is managed: prioritized, mitigated, priced, and reported. Unquantified risk, on the other hand, can only be suffered.

The endgame: finance sees machines the way it sees cash

Where all of this is heading is a digital twin of asset health: a live, continuously updated model of the physical operation and the financial exposure attached to it. Treasury has had real-time visibility into cash for years, and there's no longer a structural reason finance shouldn't have the same visibility into the equipment that generates the cash.

That's the ambition behind everything we build at Tractian. We aren't just monitoring machines. We're connecting the plant floor to the P&L.

Picture the monthly business review where, alongside cash position and pipeline, there's a third panel: asset risk. Which sites carry elevated failure probability, what revenue is exposed, what's being done about it and what it costs. Not a maintenance report translated for executives, but a financial view of operational risk, native to the language the executive team already speaks.

I believe the companies that outperform over the next decade will be the ones that got there first, and not because their machines are better, but because their visibility is.

Evaluate reliability through risk, cost, and revenue, not plant KPIs

If this conversation reaches your executive team as a debate about OEE targets or mean-time-between-failures, it will die in the room. Those are operational instruments, and while they matter enormously to the people running plants, they're the wrong altitude for an executive decision.

The executive frame comes down to three questions.

What revenue depends on assets we currently can't see into?
What is operational instability actually costing us, added up across the whole P&L rather than just the maintenance line?
What would we be willing to commit to, in customers, growth, and expansion, if we trusted our capacity the way we trust our cash?

Don't take my word for it

Verdantix, an independent research and advisory firm with more than fifteen years covering industrial and operational technology, recently built a financial model for a mid-sized manufacturer with $200M in revenue and twelve sites, validated through interviews with Tractian customers. Their findings: a 401% return over three years, $5.14M in total benefits, break-even in eight months, and roughly $1.4M in annual savings beginning in year one.

Their survey work tells you where the market already is. Of sixty maintenance, operations, and reliability leaders at firms above $250M in revenue, 98% expect to increase condition-monitoring investment through 2026. Your competitors have done this math. And it matches what we see every day with manufacturers like Ingredion, Whirlpool, Bosch, and Unilever: when reliability data reaches the finance conversation, the investment case makes itself. The full study is worth putting in front of your finance team, because it speaks their language.

View the full Verdantix study here.

One question for your next executive review

If you take a single thing from this, make it a question: "What is unplanned downtime actually costing us, in total, across the P&L, and how do we know?"

If the answer is a confident number, your organization is ahead of most. If the answer is an estimate, or a silence, you've found the blind spot, and it isn't a maintenance problem so much as a line item your balance sheet simply hasn't met yet. For the first time, introducing them is a choice you get to make.

This stopped being a maintenance conversation a while ago. It's a business conversation now, and it deserves a seat at the table. Bring it to your team, and let's get to work.

More from David Walters (1 articles)