The spreadsheet earned its job
Nobody sets out to build a system on a spreadsheet. It starts as one person's list: jobs this week, who ordered what, which invoice is still open. It is fast, free, and honest about what it knows. For a one-person operation it is genuinely the right tool, and anyone who tells you otherwise on day one is selling something.
The sheet gets its second job quietly. A colleague gets edit access. A tab appears for prices, then one for stock, then a pivot someone built in 2023 and nobody dares touch. The file is no longer a list. It is the place the business keeps its memory — and it was never designed to be that.
Watch it happen on any Friday: two copies of the same file, one called FINAL, one called FINAL-v2, both edited this week, nobody certain which one the quote went out from. That moment of doubt is the earliest honest signal — and it costs nothing to heed yet.
The failure modes are structural
When a spreadsheet breaks a business, it is almost never a formula error. It is two people editing at once and one version winning. It is a row deleted in March that nobody misses until a client calls in June. It is the question "who changed this price" having no answer, because the sheet remembers values, not decisions.
None of these are discipline problems. They are what a file does when you ask it to behave like a system: no record of who did what, no rule that stops an invalid entry, no way for two people to work without collision. The tool is not failing you. The job outgrew the tool's design.
Ask a harder question: when did the price last change, and who decided? In a real system that answer is one click — a record with a name and a timestamp. In the sheet it is a shrug. The shrug is fine for a shopping list. It is not fine for the number your margin depends on.
The arithmetic nobody does
You do not need an industry benchmark to price this — your own numbers are better. Take one week. Count the hours your team spends re-typing, reconciling, and hunting the current version of the same information. Multiply by what those hours cost you. Add whatever one lost order, one double-booked slot, or one mispriced quote cost you last quarter.
That number is yours, so it is honest. If it comes out to a few hundred a year, close this page and keep the sheet. If it comes out to a few thousand a month, the spreadsheet is already your most expensive employee — one that never improves, never flags its own mistakes, and walks out the day the laptop does.
An illustrative week, with your numbers swapped in: if three people each lose four hours a week to re-typing and reconciliation, that is twelve hours. At €25 an hour fully loaded, that is €300 a week — over €15,000 a year — spent moving information between boxes that should have been one record. Your figures will differ; the shape rarely does.
When to keep the sheet
Keep it when one person owns it end to end. Keep it when the worst outcome of a bad row is an awkward afternoon. Keep it when the business could lose the file tomorrow and rebuild from memory in a day. Plenty of healthy firms run a long time on exactly that, and the discipline of a small operation is a real asset, not a gap.
Software too early is its own waste: licences, setup, and a process frozen before you know what the process is. The goal is not to own software. The goal is that the information the business runs on has one version, an owner, and a memory of what changed.
And if the whole business still fits in one head and one file, software can wait. The honest sequence is process first, tools second: a clean weekly rhythm around one file beats a badly fitted platform every time.
When to replace it
The triggers are specific. More than one person edits the same file daily. A mistake in it has a price tag — a missed booking, a wrong delivery, a quote sent with last month's costs. Month-end means an evening of archaeology before the numbers can be trusted. New staff learn the sheet by oral tradition because the logic lives in the owner's head.
Any one of these says the sheet is now a system of record. Two or more say it is an unaudited system of record carrying real money. That is the line. Past it, the question is not whether to replace the spreadsheet but what the replacement must refuse to allow: no double entries, no silent edits, no version roulette.
There is a quieter trigger too: the moment you hesitate to give a new hire edit access, because they might break something they cannot see. A system that cannot be handed over safely is not a system — it is a dependency on one person's caution.
What replacing it actually involves
Done properly, it is boring on purpose. Someone maps how the work actually flows — not the org chart version, the Tuesday-afternoon version. The spreadsheet's real jobs get listed one by one, and each becomes a small, explicit piece of the replacement: the booking rule that cannot collide, the price table that keeps its history, the dashboard that reads the same record everyone else sees.
What it should never be: a platform subscription whose shape you rent and bend to, or a year-long project with a demo at the end. A fixed scope, written down, priced once, built against your map, delivered with the proof that it works — that is the entire trick. The sheet retires when the replacement takes its first real Monday without anyone opening the old file.
That is also how we work when a firm brings us a spreadsheet that has become load-bearing: written intake first, a map of the actual workflow, a fixed-scope plan you can read before anything is built. If the honest answer is that you do not need us yet, that is the answer you get in writing.
The first Monday after a good replacement is unglamorous: the same work, the same people, the same customers — but nobody asks which file is current, and nobody rebuilds a number before trusting it. That quiet is what you bought.
The one-line test
Ask it tonight: if this file vanished tomorrow, what would the business lose — a list, or its memory? A list you can retype. A memory you cannot.
If the answer is memory, write down what the sheet actually does for you — the three jobs it really carries — and send us that. You will get a written answer within one business day: whether to keep it, fix the process around it, or replace it, and what replacing it honestly costs. No call required, no obligation attached.
The short version
A spreadsheet fails a business structurally, not gradually: no audit trail, no concurrency, no rules. Keep it while one person owns it and a bad row costs an afternoon; replace it when more than one person edits it, a mistake has a price, and month-end means archaeology. The replacement is not a subscription to bend into — it is a small fixed-scope build that refuses the failure modes the sheet allowed. If your sheet vanished tomorrow and the business would lose its memory, that is the signal. Write down the three jobs it carries and ask in writing — the honest answer, including "not yet", comes back within one business day.