Inventory management in Excel does not break because the file gets too big. It breaks because two people open it at once, because you cannot answer what stock was on 14 March, and because a spreadsheet has no idea that 40 of those 120 units are already promised to a customer. Row count is almost never the problem.
That matters, because most advice on this subject tells small businesses to migrate when they hit some number of products or orders. That threshold does not exist. Below are the four conditions that actually decide it, what the spreadsheet costs while you wait, and why the fix is often a module you already pay for.
When does inventory management in Excel actually break?
Excel is genuinely good at stock control under three conditions: one person owns the file, stock changes at a pace a human can keep up with, and nobody else needs the same number at the same moment. Plenty of firms run for years inside those limits and should not be talked out of it.
You leave those limits the day any one of the following becomes true — and it is usually the first one:
- Two people need to write to it. The moment sales and the warehouse both edit stock, you have two versions and no way to merge them. Shared workbooks and cloud co-authoring reduce the pain; they do not solve reservation conflicts.
- You need history. A spreadsheet stores the current state. Ask what stock was on a date, or who changed a figure and when, and the honest answer is that the information was overwritten.
- You need to reserve stock. Physical stock and available stock are different numbers. A spreadsheet knows one of them, which is how you sell the same pallet twice.
- A second system holds the same number. As soon as your webshop, your bookkeeping package or a supplier portal also tracks stock, the spreadsheet becomes a fourth opinion rather than the source of truth.
What does the spreadsheet actually cost you?
Vague warnings about "inefficiency" do not help anyone decide. So here is arithmetic you can substitute your own numbers into. Assume a wholesaler with 900 active articles, three people who touch stock, 60 outbound orders a day, and roughly 225 working days a year.
- Reconciliation time. If two people spend 45 minutes a day agreeing what the real figures are, that is 7.5 hours a week, about 340 hours across those 225 days. At a loaded internal cost of €45 an hour that is around €15,000 a year in a task that produces nothing.
- Stockouts you did not need. If one order in 200 goes wrong because available stock was overstated — a sale of something already reserved — that is about 68 incidents a year on 13,500 orders. Cost per incident is rework, a call, sometimes express freight; at €60 that is roughly €4,100.
- Dead capital. Without reliable history you cannot see which articles have not moved in a year, so nobody writes them down or stops reordering them. On €200,000 of stock, even 8% quietly obsolete is €16,000 of cash sitting on a shelf.
None of those three numbers is a software feature you are missing. They are the cost of not having a single, timestamped record — which is the entire thing a stock system sells you. Whether they add up to more than a subscription is a calculation you can now do in ten minutes with your own figures.
You may already own the fix
Before comparing new packages, check what is switched off in the software you already pay for. The Dutch SME back office is unusually well served here: Exact Online, AFAS and e-Boekhouden all have stock or article modules that many customers never activate, and most webshop platforms track stock natively. A module you already licence, holding the same article codes your invoices already use, beats a better standalone tool that needs a new integration. If you do end up connecting systems, that is a koppeling question, not a migration.
If none of them fits, the next step up is a dedicated inventory management system, and above that — once the building itself is the constraint and you are managing locations, picking routes and batches — a warehouse management system. Those are three different price brackets, and skipping straight to the third is the most common overbuy.
Which three stock figures should you be able to see?
Most spreadsheet setups track one number and call it "stock". A system that is worth paying for tracks three, and the gap between them is where the expensive mistakes live.
- Physical stock. What is on the shelf right now. This is the number a spreadsheet holds, and the only one you can verify by walking into the warehouse and counting.
- Available stock. Physical stock minus what is already reserved for confirmed orders. This is the number sales should be quoting from, and it is almost always lower than the one they see.
- Economic stock. Available stock plus what is on order from suppliers and due before you need it. This is the number purchasing should be reordering from, and using the wrong one here is what produces both stockouts and over-ordering in the same quarter.
If your team currently works from one figure, that single change — separating available from physical — usually delivers more than any other feature on the vendor's list. It is also a useful test when you compare packages: ask which of the three each screen is showing, and how a user knows. Some cheap tools show physical stock everywhere and quietly leave you exactly where the spreadsheet did. Vocabulary varies between packages — and on our own inventory page the third figure is called “vrij beschikbaar” rather than economic stock — so agree which definition you mean before comparing two screens. The same distinction underpins every ERP inventory module, which is why the vocabulary is worth learning before the demo.
What to check before you migrate
Four things, and none of them is a feature comparison. Get these wrong and the new system inherits the old problem with a monthly fee attached.
- Are your article codes actually unique? Duplicates, trailing spaces and three spellings of the same supplier are the standard finding. This is the work, and it is worth doing in the spreadsheet before anything moves.
- Is your opening stock true? Migrate on a counted figure, not a believed one. A system started from wrong numbers is wrong faster and more confidently than the spreadsheet was.
- Who decides what a figure means? One named person who can settle whether goods in transit count as stock. Without that, every system disagreement becomes a meeting.
- What breaks if the internet is out? A warehouse that cannot dispatch because a cloud tool is down is a real risk in a way a local spreadsheet never was. Ask the vendor and write the fallback down.
How do you move without turning it into a project?
Run the spreadsheet and the system in parallel for one full month, then stop maintaining the spreadsheet on a date you choose in advance rather than a date it collapses on you. Migrate one product group first — ideally your fastest-moving one, because errors surface within days instead of at the next stocktake. Keep the old file read-only afterwards; you will want it, and nobody should be able to edit it.
Resist the temptation to redesign your warehouse at the same time. Layout, zoning and picking routes are a separate exercise with its own payback, covered in how to lay out a warehouse. Changing your system and your physical process in the same month means that when something goes wrong, you will not know which decision caused it.
Where AI fits — and where it does not
Nothing above is an AI problem, and it is worth saying plainly: a language model cannot fix a spreadsheet that two people are editing. What AI does well arrives afterwards, once there is a clean, timestamped record to work from — demand patterns that suggest reorder points, supplier lead times that have quietly drifted, an article whose margin has moved without anyone noticing. That is also why the sequence matters. Get the record right first; the analysis is cheap once the data is true, and expensive guesswork until then. When you are ready to cost that step, we publish figures for what an AI project costs.
Last updated 19 August 2026.
