There is a spreadsheet somewhere in your business doing real work. It tracks jobs, or pricing, or inventory, or scheduling. It has grown for years. It has tabs nobody opens and formulas nobody has looked at since 2021.
One person built it. That person is the only one who genuinely understands it.
That spreadsheet is simultaneously the most valuable and most dangerous object in the company.
Why it exists
Not because anyone was careless. Because it was the fastest thing that worked.
Somebody had a problem, opened a spreadsheet, solved it in an afternoon, and it kept working. Then it absorbed another problem. Then another. Nobody ever sat down and decided this would be the system of record. It just became one, one column at a time.
That is a completely rational path and it is how most small businesses actually run.
Why it is a problem
It has one maintainer. If that person is on holiday, ill, or leaves, the business has a capability it can no longer exercise. Nobody else can safely change it, because nobody knows what the change would break.
It fails silently. A broken formula does not throw an error. It produces a wrong number that looks exactly like a right number, and it can do that for months.
It does not scale. There is a size where it becomes slow, fragile and unmanageable, and the transition is usually a crisis rather than a decision.
Nobody trusts it, quietly. People start keeping their own version. Now there are three versions, all slightly different, all being used for decisions.
It cannot connect to anything. The rest of your stack cannot read it, which means humans become the integration layer, retyping between systems.
What not to do
The reflex is to replace it with software. Sometimes right, often a mistake.
The spreadsheet works because it fits your business exactly. It grew around your actual process. Off-the-shelf software fits somebody's average customer, and the gap between the two is where adoption dies. See our note on why your team ignores the CRM.
Do not rip it out until you understand what it is genuinely doing.
What to do instead
First, document it. Not the formulas — the purpose. What decisions depend on this? Who enters what, when? What does each tab actually do? What are the known quirks and workarounds?
Sit with the person who built it and record them explaining it. An hour of that is worth more than any amount of reverse-engineering later.
Second, remove the single point of failure. At minimum, a second person who can operate it. Better, a written procedure. Our note on writing your first SOP covers the fast version.
Third, protect the data. Version history on. Backups that actually exist. Not "it is in the cloud" — a copy that survives someone deleting a tab.
Fourth, identify what is actually fragile. Usually one or two things: a manual step that gets forgotten, a formula range that breaks when rows are added, a lookup that fails silently on a typo. Fix those specifically.
Fifth, then decide about replacing it. Now you know what it does and what it is worth, you can evaluate honestly. Sometimes the answer is a small piece of custom tooling that does the one hard part and leaves the rest alone. Sometimes it is a real system. Sometimes it is: leave it, it works, just make it safe.
The middle path people miss
The most common right answer is not "keep the spreadsheet" or "buy software." It is to take the two or three genuinely automatable steps out of it — the data entry, the recalculation, the report that gets built by hand every Monday — and leave the rest.
That removes most of the risk and most of the labour without a migration project, and the business keeps the thing that actually fits it.
The test worth running
Ask someone who did not build it to produce this week's numbers from it, alone, with no help.
Whatever goes wrong in the next twenty minutes is your actual risk register.