Spare parts management: what to stock and what not to

Two opposite problems coexist in the maintenance warehouse, almost always at the same time: parts that are never there when needed, and parts that have been sitting untouched for two years tying up money on a shelf.

Both come from the same place: the list of what to stock was decided once, by eye, and never revisited. This article is about deciding it with data instead.

What’s worth keeping in stock

The question isn’t which parts you use, but which ones stop a job if they’re missing. There are far fewer of those than it seems, and the criterion has three parts:

Consequence. If it’s missing, does the visit stop halfway and require a return trip? A second visit gets paid for twice: the travel and the hour.

Lead time. If the supplier can deliver it tomorrow, you probably don’t need to stock it. If it takes three weeks, you do.

Frequency. Something consumed several times a year, yes. Something used once in five years, almost never.

With those three criteria, the list usually comes down to fewer than a hundred items, and those are the ones that need a minimum stock level in warehouses and items.

And what’s not worth tracking

Just as important. There’s an item flag for no stock tracking, for consumables that aren’t worth counting.

A system that forces you to inventory screws, cable ties, and electrical tape gets abandoned within two weeks, and drags down the tracking of what actually mattered along with it. It pays to be deliberately selective.

The van is a warehouse

The change that reconciles inventory fastest, and the one fewest companies make.

In an operation with field technicians, a large part of the material lives in the vehicle. As long as that isn’t in the system, the system’s stock and the real stock will always be two different things, and nobody will know whether a part is available or already installed.

A warehouse can be linked to a technician team, a client, an address, or a supplier. And transfers between warehouses record who requested them and who approved them: that double sign-off turns a movement into reliable data.

Consumption isn’t a separate entry

The idea that everything else rests on: consumption is a consequence of closing the order, not a follow-up administrative task.

The technician consumes the item straight from the work order in the app, and with that action stock goes down and the cost of the intervention goes up. It works without connectivity and syncs afterward.

If consumption gets logged on Friday, stock has been inaccurate all week and the cost of closed orders is incomplete.

The original reference number

A small field that solves a big problem. Each item carries, besides its code and cost and sale price, its original manufacturer reference, its brand, and its model.

That’s what lets you find another manufacturer’s equivalent when the original part has a lead time of weeks. In automotive, HVAC, and industrial machinery, that’s the difference between a two-day stoppage and a two-week one.

What expires

Not all spare parts last forever: consumables, liquids, adhesives, batteries, refrigerants, and much of the electrical material carry a date.

Items that need it are managed by batch, with its code, quantity, manufacturing date, and expiration date, and with its own cost and price per batch. And batches are located in warehouses, so you know which specific batch is where.

In sectors with strict traceability requirements, this also lets you answer which batch was installed in which piece of equipment, and on what date.

How to know what’s surplus

This is where the history pays off.

Reports on materials used in work orders, movements, and inventory show which items actually move and which have been sitting still.

And cross-referencing that with orders marked as waiting on parts shows the opposite: what’s causing return visits and, therefore, what should have a minimum stock level and doesn’t.

That cross-check, reviewed once a year, tends to reshuffle the initial list quite a bit.

When to buy

What prevents the expensive rush order isn’t the purchasing module: it’s having minimum stock levels correctly set on the items that stop work. The purchasing workflow handles what comes after, with its supplier, destination warehouse, and line items.

You can also define a per-supplier price for each item, with its agreed discount and reference number, so the agreed percentage applies automatically when generating a purchase order and delivery notes no longer need to be checked line by line.

If there’s an ERP involved

The rule is agreed before connecting anything: the ERP governs items and pricing; the CMMS governs consumption. Items carry their ERP identifier precisely for that purpose.

Integrations documents the scope of each connector, which isn’t the same across the board.

How stock on hand is valued

A decision almost nobody asks about in a demo, and one that determines the cost you’ll see on every work order: the warehouse’s valuation method.

In GMAO Cloud there are four: weighted average cost (the default), actual cost, FIFO, and LIFO. And it’s set per warehouse, not for the whole company.

It matters because the purchase price of the same item changes between orders. With average cost, the cost of an intervention is smoothed out and easier to compare across periods; with actual cost or FIFO, it reflects what that specific part truly cost.

Neither is better in the abstract: it depends on whether you need to compare or to allocate exactly. What is a mistake is not knowing which one you’re using, because then the cost per asset shown in reports doesn’t mean what you think it does.

The physical count, at the end

The full count is the job everyone puts off, and for good reason: it takes days and, two weeks later, it’s out of sync again if consumption keeps getting logged late.

The order that works is the reverse. First fix the recording — consumption on order close, vans registered as warehouses, transfers with approval — and only then count. That way the inventory you take stays accurate, instead of being a snapshot that goes stale from day one.

If you’d like to see this applied to your own items, you can request a demo.