Every established business has one. Sometimes several.
It started as a way to keep track of something for a fortnight. It is now eleven tabs, has formulas referencing two other workbooks, drives a decision worth real money every week and is understood by exactly one person.
Nobody planned this. That is rather the point. Nothing that important gets deliberately built in a spreadsheet. It arrives there one reasonable extension at a time. The moment it stopped being a spreadsheet and became a system passed without anyone noticing.
It is not the spreadsheet's fault
Worth saying plainly, because the instinct is to treat this as a failure of discipline. It is not. Excel is a genuinely remarkable tool and the person who built the thing was solving a real problem quickly with what they had. In most cases they saved the business a great deal of money doing it.
The problem is not that it exists. The problem is that it kept being given more to do.
Five tests
Not all of these are urgent. Three or more together generally is.
One person can explain it. The real test is not whether somebody else could work it out. It is what happens if they have to work it out in a hurry, in the week that person is on leave, with a client waiting. If the honest answer is "we would have a bad week", that is key-person risk wearing a green tick.
It has become the record. There is a difference between a workbook you use to calculate something and a workbook that is where the answer lives. Once the spreadsheet holds the only copy of something, whether that is the current price list, the stage every job is at or who has been invoiced, it has become a database with no backup discipline, no audit trail and no controls on who changes what.
Two people cannot use it at once. Someone has it open. Someone else needs it. The workaround is a copy. Now there are two versions and in about a fortnight nobody will be certain which one is current. This is the point at which a spreadsheet actively starts costing you money rather than saving it.
Nobody can say when a number changed. Something is wrong in a figure a client has already seen. When did it change? Who changed it? What was it before? A spreadsheet cannot tell you. That is fine for a working calculation and not fine for anything with a compliance or contractual dimension.
Fixing it requires courage. There is a formula nobody will touch. It works. It is nine lines long. The person who wrote it is not certain any more either. Everybody has agreed, without discussing it, to build around it rather than into it. That agreement is the clearest signal you will get.
What actually happens next
The reflexive response is to buy a system. Sometimes that is right. More often it is expensive and disappointing in a specific way: the package does ninety per cent of what the spreadsheet does, does it differently and the remaining ten per cent turns out to be the ten per cent that made your business work. So the spreadsheet survives alongside the new system, which is the worst of both.
The alternatives are worth looking at honestly first.
Sometimes it is a settings problem. The capability exists in software you already pay for and nobody turned it on. This is a genuinely common outcome and the cheapest one on the list.
Sometimes the process should change. The workbook has grown to accommodate exceptions that a decision could remove. Fewer exceptions, smaller problem, possibly no problem.
Sometimes it should be built. When the process is genuinely specific to how you operate, building the thing that fits is now affordable in a way it was not five years ago. That is more often the case in trades, logistics and professional services than software vendors like to admit. AI changed the economics of small custom applications considerably.
The part people get wrong
Replacing it all at once, at the weekend, with a cutover.
The reason it is eleven tabs is that eleven things needed doing. You will not discover all eleven by reading it. You discover them by moving one at a time and finding out what breaks, while the spreadsheet is still there to fall back on.
Slices, not a cutover. The workbook shrinks tab by tab until the day somebody notices nobody has opened it in a month.