The spreadsheet you should replace first

Almost every small business runs on spreadsheets, and most of them should keep doing so. A spreadsheet that one person maintains, that produces a number someone reads once a month, is a perfectly good piece of software. Replacing it buys you nothing.

The one to replace is the one being edited by two people at once.

Why is concurrent editing the tell?

Because it fails quietly. When two people work the same file, one of them overwrites the other, and nobody gets an error. The wrong number just sits there looking exactly as authoritative as the right one.

Everything else that goes wrong with spreadsheets is loud. A broken formula shows #REF!. A missing file is missing. But a stock count that was 40 and is now 12 because someone had an old copy open looks completely normal, until you promise a customer something you do not have.

If you find yourself saying “let me know when you’re out of it so I can go in”, you are paying a coordination tax on something a database solves outright.

What else makes a spreadsheet worth replacing?

In rough order of how much it costs you:

• Two or more people edit it in the same week. As above. This is the big one.
• It feeds a decision about money. Pricing, margin, payroll, stock. Errors here have a direct cost, so correctness is worth paying for.
• Somebody re-types data from it into something else. Every manual re-entry is a guaranteed future error and a job nobody wants.
• It has a version in its filename. _v4_FINAL_updated means the file is not the record: someone’s memory of which file is current is the record.
• Only one person understands it. A 300-row nested formula is a dependency on a human being. When they leave, the process leaves.

Which ones should you leave alone?

Genuinely most of them. Leave a spreadsheet alone when:

• One person owns it and always will
• It is a calculation, not a record: a model, a forecast, a one-off analysis
• It changes shape often, because the flexibility is the point
• It works and nobody complains

There is a real temptation, once you start building software, to replace everything. Resist it. Purpose-built software is better than a spreadsheet at being a shared source of truth, and worse at letting one person restructure their thinking on a Tuesday afternoon.

What does replacing it actually look like?

Not a big system. The first version should do less than the spreadsheet does.

Pick the part where the concurrency hurts, usually one table that several people touch, and move only that. Keep the spreadsheet for the analysis on top of it, pointing at exported data. You get the correctness where correctness matters and you keep the flexibility where flexibility matters.

That is usually a small piece of work rather than a project, and it is a good first thing to build with anyone: small enough to be finished, real enough that you find out how they work.

How do you know it worked?

Nobody asks who has the file open. That is the whole measure.