Read This If You've Had a Large MRO Inventory Write-Off
A CFO's guide to why it happens, and how to make sure it doesn't happen again
If your company just took a large write-off on maintenance, repair, and operations (MRO) inventory, you're not alone. It happens at almost every asset-intensive manufacturer, eventually. The good news: it's predictable, and it's preventable.
Why This Keeps Happening
MRO inventory doesn't behave like normal inventory. Production parts flow. MRO parts sit (sometimes for years) until a machine breaks down and pulls them off the shelf. That difference is the root of the problem.
Here are the usual suspects behind a write-off:
No owner. Procurement is measured on cost. Maintenance is measured on uptime. Each side hedges against the other, operations over-requests “just in case,” procurement over-orders to avoid stockouts. Nobody owns the working-capital impact.
Duplicate parts, everywhere. Plants create their own item numbers. The same bearing ends up with five different part numbers across five plants. Safety stock is duplicated and excess inventory builds.
No process for setting the right stock levels for new equipment. When a new machine is purchased, nobody stops to ask which spares it needs, how critical it is, or how much stock to carry. Reorder points get set on a guess and that guess never gets revisited.
No visibility or multi-echelon planning. One site orders a part that's sitting idle at a sister plant three states away. No one is responsible for planning where common spare parts should sit or how the sites should work together to maximize uptime with lower inventory levels.
No process for retiring equipment. When a machine is decommissioned, its spare parts are supposed to go with it. Usually, nobody plans the phase out of spare parts. The parts just sit there, quietly becoming dead stock.
None of this is unusual. Industry research suggests 30 to 50 percent of MRO inventory at a typical manufacturer hasn't moved in two years or more. Most companies are carrying the wrong version of a very large number: too much of what rarely fails, too little of what actually stops the line.
The Real Cost Isn't Just the Write-Off
The write-off is the visible cost. The hidden cost is worse: while you're overstocked on parts that don't matter, you're often understocked on the ones that do. Unplanned downtime from a missing critical spare can run into hundreds of thousands of dollars per hour, and a two-hour repair can turn into a multi-day outage waiting on emergency freight.
So the goal isn't just “reduce inventory.” It's “hold the right inventory, in the right place, for the right reason”.
What Actually Fixes It
A one-time cleanup doesn't work. Six months later, the same duplicates and dead stock come right back. Lasting fixes need the following things working together:
A named owner, with the right incentives. Someone has to own the balance between inventory cost and uptime — today, usually nobody does. Procurement is measured on spend, maintenance is measured on uptime, and each hedges against the other. Put one accountable owner in place, and change how that role and the people around it are measured, so the incentive is to get the balance right, not to protect one side of it.
Implement process change, backed by real change management. New equipment purchases need a step to set stocking levels up front. Retiring equipment needs a step to release its spares. Buyers and planners need to work differently than they have for years. None of this sticks without training, clear new policies, and deliberate change management — otherwise old habits quietly take back over within a few months.
Use real demand science. Replace static min/max with models built for intermittent, failure-driven demand, not steady sales-style demand. Segment parts by value, predictability, and how critical they are to keep running.
Clean, de-duplicated data. Use AI to match parts across plants — even when descriptions don't match — so you know what you actually have and where.
An AI/machine-learning layer on top of the ERP. Your ERP wasn't built to plan for parts that fail twice in five years and then not at all. Layer AI and machine-learning models on top of it to read intermittent, failure-driven usage patterns, and generate the reorder points and safety-stock levels the ERP alone can't produce.
Skip any one of these, and the problem comes back.
How This Plays Out: A Case Example
A multi-plant automotive supplier came to this exact point. Ten plants, running on a common ERP, no shared visibility between them. MRO inventory had grown to almost $100 million and nearly 30 percent of it hadn't been used in over four years. Another $30 million MRO writeoff was coming.
The fix started small: two pilot plants, twelve weeks. That first phase fixed the planning parameters, put in basic accountability, found the duplicate parts and built a short list of stock ready to burn down. New procedures were developed and rolled out at the pilot plants. This phase paid for itself in less than a quarter and built the foundation for the next phase.
The next phase rolled out across all ten plants — with training, new buying policies, and hands-on change management so buyers and planners actually adopted the new way of working, not just a new dashboard. Only after the data was clean did they tackle multi-echelon inventory and supplier consolidation.
Each phase funded the next. No single phase required a leap of faith.
What This Means for You
If you've just absorbed a write-off, the instinct is to order a cleanup and move on. Resist that. A cleanup alone will be back on your desk in two years as another writeoff.
Instead, ask three questions:
Who owns MRO inventory and the balance between inventory cost and uptime today? If the honest answer is “no one,” that's the first gap.
When we buy new equipment or phase out equipment, is there a defined step to set the right stocking levels — or does that happen by accident, if at all?
Do we have the ability to pro-actively plan our MRO inventory levels and stocking locations, or do we wing-it?
If you don't have confident answers, that's where to start. The fix is well understood, it pays for itself in the first phase, and it's the difference between writing this off once and writing it off again in three years.
AI Disclosure
Writing & Industry Experience: Greg Pitstick
Editing: ChatGPT
Image: Claude