Nobody chose this. That is the first thing to say, because the conversation usually starts with somebody looking slightly embarrassed about a folder called Comenzi 2026 FINAL v3.
Nobody sat down and designed a business around shared workbooks. What happened is that a spreadsheet was exactly the right tool for the first version of a problem, and then the problem grew and the tool did not, and every individual day it was easier to add a column than to change anything. That is a rational sequence of decisions that arrives somewhere irrational.
So this is not an argument that spreadsheets are bad. It is an attempt to price what yours is costing, because the number is invisible by design.
Where the money actually goes
Re-typing. The same order, entered into the sheet, then the ERP, then the delivery note. Two minutes each, forty times a day, is eighty minutes. Call it 350 hours a year of pure transcription, at whatever your loaded hourly cost is.
Reconciling. Two people edited the file. One had it open since morning. Somebody now spends an hour working out which version is right, and the answer is usually "neither, entirely".
Asking. The single most expensive activity in an operations team is a person interrupting another person to ask a question the system should have answered. Where is that order. Did we invoice this. What did we quote them last time. Every one of those is two people's attention, not one.
Waiting for the person who knows. Every spreadsheet operation has one individual who understands the formulas in columns AA through AF. When they are on holiday, the business runs at reduced capacity and everyone pretends this is normal.
The decisions you do not make. This is the largest cost and the hardest to see. When getting the number takes half a day, you stop asking for the number. Which products actually make money after returns. Which client is quietly unprofitable. Nobody decides not to know these things. They just become too expensive to ask.
The arithmetic, done honestly
Take one number: hours per week your team spends moving data between places, rather than doing anything with it. Not estimating generously. Ask three people and take the middle answer.
Say it is ten hours a week. That is roughly 500 hours a year. At 15 EUR an hour fully loaded, 7.500 EUR annually, forever, rising with headcount.
Against that, a first-generation internal system starts around 3.000 EUR and most land between 3.000 and 12.000, one-off. There is an instant estimate if you want a number for your own shape of problem before speaking to anybody.
The comparison is usually not close. What stops people is not the maths. It is that the 7.500 is spread across fifty-two weeks and twelve people and never appears on any line of any report, while the 3.000 arrives as one invoice with a date on it.
When the spreadsheet is right and you should keep it
I would rather say this plainly than have you find out by spending money.
Keep the spreadsheet when the process is genuinely still changing shape every few weeks. Software encodes decisions; if the decisions are not made, you will pay to encode them twice.
Keep it when one person does the whole thing and nobody else needs to see it. Most of the cost above is coordination cost. One person coordinating with themselves has none of it.
Keep it when it is genuinely analysis rather than operations. Excel is an outstanding analysis tool. It is a poor multi-user database, and the trouble starts precisely when a file crosses from the first thing to the second, which it does without anybody announcing it.
The tell is not the size of the file. It is the number of people who need it to be correct at the same moment.
What replacing it actually looks like
The fear is a two year project that ends in a system nobody wanted. That is a real thing that happens, and it happens when the replacement is designed from a wishlist rather than from the sheet.
What we do instead: the spreadsheet is the specification. Not a starting point for one, the actual specification. The columns your team added by hand over four years are a record of every question the previous tools could not answer, and they are more honest than any requirements document, because nobody was performing for a stakeholder when they added them.
So the first version does what the sheet does, for the people who use the sheet, and nothing else. It ships in weeks. Then the additions start, and they are cheap, because the hard part was never the software.
And the file itself survives, as an export. People still want their numbers in Excel to do things nobody anticipated. What changes is that it stops being the place the truth lives, and starts being a copy of it. That distinction is the whole project.
The version of this that has no software in it at all
If the reading above has convinced you of nothing else, do one thing this week. Write down, for a single day, every time somebody copies a value from one place to another by hand. Just tally it.
Most teams are surprised, and a few find something they can fix with a shared view and a naming convention and no budget at all. That outcome is fine. It is a better outcome than a system you did not need.
The point was never that you should buy something. It is that the current arrangement has a price, it is being paid, and almost nobody has ever added it up.