This is the most expensive question in the warehouse. Too much stock costs money, space and parts that age on the shelf. Too little costs downtime, express shipments and improvisation. The question can't be solved with a formula, but it can be solved with a trade-off you can explain. For the broader framing, see spare parts management.
The question behind the question
Holding stock is not a goal but an insurance policy. You pay a premium — capital, space, the risk of parts becoming obsolete — in return for the certainty that you won't stand still.
Framed that way the decision sharpens. You no longer ask "do we need this?", but "is the premium lower than the damage it prevents?". That is a question you can answer per part and explain to a controller.
Four factors that decide it
What downtime costs
Start here, because this weighs heaviest. A part in a machine that carries all production is a different story from the same part in a standby pump that is only there for reassurance.
Calculate in euros per hour or per day; rough is good enough. A stopped line costs production, possibly overtime to catch up, sometimes penalties or rejected material. Many organisations have never calculated this number, and that is exactly what makes the stock discussion so slippery.
Where there is redundancy the figure often drops dramatically: with two pumps where one can fail without consequence, urgency is a fraction of that for a single installation.
How long delivery takes
A part that is with you tomorrow doesn't need to be in stock. A part with a fourteen-week lead time does, even if you rarely need it.
Calculate with actual lead time, not the promised one — we cover that difference in calculating reorder point and safety stock. And look beyond the standard lead time: can the supplier expedite, and what does that cost? Sometimes an express charge of a few hundred euros is cheaper than years of holding stock.
How likely you are to need it
For wear parts you know this from history or the maintenance schedule: a filter replaced every 2,000 run hours has predictable consumption.
For failure parts it is harder. There the analysis you make in an FMECA helps: which failure modes exist, how often they occur and what the consequences are. Without that analysis, your technicians' experience is the best alternative — ask which parts they have ever had to arrange in a hurry.
What it costs to have it
Price is only part of it. Add: the space it occupies, the capital tied up, and the chance the part becomes obsolete before you use it. Electronics and rubber parts have a shelf life; cast iron parts will sit there just the same in ten years.
The decision tree
The first three factors form a decision tree; the fourth then determines how many you hold. That way most parts choose their own place.
The outcome isn't always black and white. Between "hold stock" and "order when needed" sit intermediate forms: an agreement with the supplier to reserve a part for you, sharing with a nearby company, or a framework contract with a guaranteed lead time. Those cost less than own stock and cover more than doing nothing.
Insurance parts
Some parts fall outside any calculation: expensive, rarely needed, long lead time, and when they fail everything stops. A main drive, a large frequency converter, a specific casting.
These are called insurance parts for a reason. You don't buy them because the chance is high, but because the consequence is unacceptable. That is a management decision, not a warehouse decision — put the trade-off forward with the downtime figure attached and have someone with authority sign it.
What you do need to arrange for these parts: preservation, periodic checks, and recording that they exist. An insurance part that turns out to be untraceable or unusable after eight years has been paid for twice.
What you don't need in stock
The reverse list matters just as much. Standard fasteners and parts any technical wholesaler delivers tomorrow don't need to be in your warehouse. Neither do parts for machines you'll decommission within a year.
Be strict here, because this category grows by itself. Every part ever ordered "just in case" and never used will still be there in five years. See also your ABC analysis: C items without consumption are the first candidates for disposal.
Record it and revise it
Record per part why it is in stock, not just that it is. One line is enough: critical for line 3, lead time 12 weeks. Without that reason the decision becomes untouchable after a few years, because nobody remembers why it was made.
Revise the list whenever something changes in your installation. A new machine, a changed lead time or a decommissioned line are all reasons to look again.
Frequently asked questions
What percentage of my parts should be in stock?
There is no norm, and any percentage you read somewhere comes from a different context. What counts: can you explain per part why it is there? If not, the stock has arisen rather than been chosen.
What if I don't know the downtime figure?
Make a rough estimate with production. Even a figure with a fifty percent margin makes the discussion more concrete than no figure at all. You don't need precision, you need to distinguish orders of magnitude.
My supplier says he can deliver anything quickly. Can I stop holding stock?
Only if you have measured that promise. Ask for realised lead times over the past year and record what happens if he misses. A promise without an agreement is not cover.
How do I handle parts for machines about to be replaced?
Wind down, but not to zero. Keep the parts whose failure would delay the replacement, and dispose of the rest. Record the machine's end date on the item so the decision explains itself.