When should a spreadsheet become custom software?

Every operations team has one: the spreadsheet with dozens of tabs, cross-references, and formulas that only its creator can safely touch. When that person is on leave, changes wait. When they eventually move on, the knowledge leaves with them.

We once replaced a 40-tab dispatch spreadsheet at a logistics company where exactly this had happened. The file worked, but every new order touched multiple tabs, and mistakes surfaced days later, when they were expensive to unwind. The spreadsheet wasn't the problem. The problem was that a business-critical workflow had a bus factor of one.

If someone on your team spends part of every day moving data between the spreadsheet and other systems (the accounting tool, the CRM, an email inbox), you're paying a salary for data transport. Worse, every manual hop is a place where errors enter silently.

This is usually the cheapest problem to fix, because the fix doesn't have to be a big system. Sometimes it's a small integration that syncs two tools. Sometimes it's an automation that reads the inbox. The point is to make the computer do the copying.

A spreadsheet that multiple people edit concurrently drifts. Formulas get overwritten, rows get sorted incorrectly, someone pastes values over a calculation. Eventually a meeting happens where two people bring two different versions of the truth, and everyone quietly stops trusting the report.

When decisions are being made on gut feel because the data 'is probably wrong anyway', the spreadsheet is no longer saving you money. It's costing you decisions.

A process that worked at 10 customers often breaks at 200, not dramatically, but through accumulating workarounds. A second spreadsheet appears to cover what the first one can't. Then a shared inbox. Then a checklist that explains how to use the other three things.

Growth doesn't just add volume; it adds people who weren't there when the process was invented. Every workaround that lived in someone's head now has to be taught, and every teaching is lossy.

Replacing a spreadsheet with custom software isn't about making things look more professional. It buys you specific, concrete properties:

• Validation at the point of entry: bad data gets rejected when it's typed, not discovered at month-end.
• One source of truth: everyone sees the same numbers, updated in real time.
• Permissions: the intern can update order status without being able to break the pricing formula.
• History: you can see who changed what and when, and undo it.
• Integrations: data flows to and from your other tools without a human ferrying it.

Honestly: often. If the process is still changing shape every month, a spreadsheet's flexibility is worth more than software's structure. If only one or two people use it and the data isn't feeding decisions elsewhere, the risk is contained. And if the pain is mild annoyance rather than real errors or lost time, the build won't pay for itself.

The right moment is usually when the process has stabilized but the volume, the team, or the stakes have outgrown the tool. That's when structure stops being a constraint and starts being a relief.