Spare parts inventory: decide which parts to hold and which to order on demand.
For large workshops, dealerships, parts distributors and plant maintenance stores. Thousands of items, most of which have not moved in months, and a few that stop a vehicle or a machine if they are missing. The system separates one kind from the other using your history of issues and proposes the week's order with the reason.
Book a 30-minute demoHow it works
Read-only connection · Tested on your own data first · Data hosted in the EU · Demos in English or Spanish
The short answer
Spare parts inventory optimisation software classifies every item by how it really moves: those that turn over, those that sell now and then and those untouched for over a year. It suits dealerships, large workshops, parts distributors and plant maintenance stores. It calculates the order for fast movers, the risk for sporadic ones and the money tied up in the rest, merges equivalent part numbers into one history and gives critical parts their own service level.
By Manuel Gómez González, founder of Arume · Reviewed on
What it does
Separates what moves from what is parked
In a typical parts store, more than half of the items have not been issued in the past year. The system classifies each by what actually leaves, how many times and how often, not by catalogue value: the brake pad that sells every week, the filter that sells every two months and the radiator that has sat fourteen months on the shelf.
Critical parts with their own service level
A bearing that stops a production line is not measured like a trim strip. You mark critical items (or we take them from the maintenance plan) and the system sets a higher service level for them and watches them separately, warning you when the risk of running out passes the level you set, even if stock still looks comfortable.
Lumpy demand, treated as what it is
An item that sells three in March, none until July and two in September cannot be forecast with an average. The system uses a method designed for intermittent demand: it estimates when the next sale will come and how many, as a range of probabilities, and that gives the real stock-out risk.
Equivalents and superseded part numbers
The same part under three numbers (original, aftermarket equivalent, and the one the manufacturer renumbered) is three short sales histories that are really one. We merge them in the first review so the system does not learn that none of the three sells, and so the order goes to the one that pays.
Orders with each supplier's lead time
The dealership delivers in 24 hours, the national distributor in three days and the importer in six weeks. Each item carries its supplier and lead time, and the quantity covers that lead time at the service level you set. A part with a fast supplier hardly needs stock; one that takes six weeks is flagged before you notice.
Dead stock in money, not units
The list of items untouched for more than twelve months, with their cost value, ordered by money so you start with what hurts. And the opposite list: parts you have rush-ordered three times in the last quarter that still have no minimum.
Why is spare parts about what to hold, not how much to order?
In a grocery shop almost everything turns over. In a parts warehouse the opposite is true: of 15,000 items, perhaps 2,000 sell regularly, another 4,000 sell now and then during the year and the rest has not moved in over twelve months. That is the usual picture in a dealership, a parts distributor or a factory maintenance store. The money tied up is not in the 2,000 that turn: it is in the 9,000 that do not.
The question that decides the warehouse is not "how many brake pads do I order?", which any ERP answers with a minimum. It is "do I need to hold this sensor, or order it when the car arrives?". Today that decision is taken from memory by a parts person with twenty years in the job, or not taken at all: the minimum someone set in 2019 is still there and the system keeps ordering.
Spare parts inventory optimisation answers that question with the history of issues. For each item it calculates how often it leaves, how many, how regularly, and what it costs to hold against what it costs not to have it. From that it proposes a group: hold with a minimum, order on demand, or phase out.
What are the three kinds of item?
The first step is to classify the warehouse by what really leaves. Not by family, catalogue price or the manufacturer's recommendation: by the history of issues of each item in your warehouse. Three groups come out, each managed differently.
Fast movers (filters, pads, oils, belts, maintenance consumables) are treated like any high-turnover product: forecast, service level and a weekly supplier order. Sporadic items (a water pump that sells five times a year, a starter motor) are forecast with a model made for lumpy demand: when it will sell next and how many, as a probability. And parked items are counted in money and put in front of whoever decides.
- Regular turnover: weekly forecast, reorder point calculated and reviewed each season.
- Sporadic demand: probability of an issue before the supplier lead time; held only if it pays or if it is critical.
- No movement in over twelve months: list in money and by age, and automatic repurchase blocked.
- Critical (marked by you): own service level, even if rarely issued, and a separate alert.
How are critical parts measured differently?
Some items are issued once every two years and still have to be held. In industrial maintenance it is obvious: a particular drive or bearing stops a line, and every hour of downtime costs more than holding the part for ten years. In a dealership it is less dramatic but just as real: the customer's car occupies a lift for three days while the part arrives.
For those items the system does not apply the "sells little, do not hold it" rule. You mark them (or they are imported from the maintenance plan or the manufacturer's list) and each gets a higher service level than the rest. They are watched separately: when the risk of running out passes the threshold, they appear in the panel with priority, even if today's stock looks like a year's supply.
What you gain is an explicit critical list with numbers instead of one in a single person's head, and the ability to discuss it: if a part marked critical has not been issued in four years and the supplier ships in 48 hours, the system shows it and you decide whether it stays on the list.
How do you handle three part numbers for one part?
It is the most common dirty data in parts. The same part exists with the original number, with an equivalent manufacturer's number and with the new one assigned when the brand changed its numbering. In the ERP they are three items with three short sales histories that, looked at separately, say none sells. Looked at together, it is a part that sells every month.
In the first review we build the equivalence table from what you have: the manufacturer's, an electronic catalogue such as TecDoc (which, according to TecAlliance, supports search by part reference and by OE number) and what the parts person knows from memory. History is merged by part and the system calculates for the part. When it proposes the order, it picks the number that pays on price and lead time and shows you the others.
The table stays in your name and grows with each new number. It is a one-off job that many warehouses have never done because there was nowhere to put it.
Which supplier, which lead time?
Parts supply mixes lead times of 24 hours (the local distributor), three or four days (the brand's central warehouse) and weeks (imports, parts for old models, made-to-order items). A single minimum per item ignores this, and the result is excess stock of what arrives tomorrow and stock-outs of what takes six weeks. Vehicles also stay on the road long: according to ACEA, the average age of cars in the EU is 12.3 years, vans 12.5 and trucks 13.9, so ordering parts for models that are no longer made is a weekly job.
The system stores each item's usual supplier and real lead time: not the promised one but the one measured from your past orders. With that lead time it calculates the quantity that covers demand until the next order can arrive, at the service level you set. Where the supplier ships in a day, stock at home drops to almost nothing; where it takes weeks, it rises and the system warns early.
If a part has two suppliers, the order goes to whichever pays that week, and if the fast one is dearer, the system puts both numbers in front of you: what the extra costs and what waiting costs.
What do you need to provide, and what does it connect to?
Issues by item and date: repair orders, sales notes to workshops or maintenance work orders, as the case may be. Current stock. Supplier receipts with order and receipt dates to measure lead times. And, if they exist, the equivalence list and the critical list. Almost every workshop or parts system exports this; what it does not, can be reconstructed. We read from Odoo, Microsoft Dynamics 365 Business Central, Sage, SAP Business One or a spreadsheet by family, with a read-only user.
Customer data, registration plates and names are not loaded. Item, date, quantity and warehouse are enough to forecast issues. Before calculating anything we send a data-quality report: items without movement, duplicate numbers, receipts without a date, lead times that do not add up. That report alone is often the first clean-up the warehouse has had in years.
How is it tested, and how do you start?
Before deciding anything live, the system replays your last months without knowing the future: what it would have held, ordered and when, compared with what you did. We measure demand served, rush orders avoided, idle money and number of orders. If it does not improve on what you do, we say so then. That is the simulation with your data.
Then one or two months in shadow mode: each week it proposes the order list and the list of items to phase out, the parts person decides as usual and notes what they accept. At the end we measure. When it convinces, it goes live connected to your ERP. To judge in advance whether this is for your warehouse, the industrial supplies page covers the neighbouring case.
Frequently asked questions
We hold over 20,000 items and most do not move. Does it work?
It is exactly the case. The system does not try to forecast all 20,000 equally: it first classifies which have enough history, which sell sporadically and which have been untouched for over a year. For the first group it calculates orders; for the second, risk and timing; for the third, how much money is tied up and what to do with it.
How does it treat a part that sells twice a year?
With a model for intermittent demand: it estimates when it will sell next and in what quantity as probabilities, not as an average. An average of 0.17 a month is no use for ordering; knowing there is a 20 percent chance it sells before the supplier can ship is. If the part is also critical, the service level goes up and it is held at home even if it rarely sells.
What counts as a critical item?
Whatever you mark as such: a part that stops a machine, a vehicle or a delivery if it is missing. You can mark them by hand or import them from the maintenance plan or the manufacturer's list. Those items get a higher service level and a separate alert, with priority over the rest.
What about equivalents and numbers the manufacturer renames?
In the first review an equivalence table is built: different numbers that are the same part. Sales history is merged and the system calculates for the part, not each code. When ordering, it proposes the number that pays on price and lead time. The table is yours and grows as new numbers arrive.
We work with several suppliers with very different lead times. Is that handled?
Each item carries its usual supplier and real lead time, measured from your past orders (what suppliers promise and what they deliver do not always match). The quantity to order covers that lead time at the service level you set. If a part has two suppliers, the system proposes the one that pays and shows you the other.
What do we do with stock that has not moved for years?
The system gives you the list in money and by age, but the decision is yours: return to the supplier if they accept, sell off, move to another warehouse in the group or keep if the part is critical. What it does is stop you buying again what is already parked, which is what happens when a minimum is badly set.