Spreadsheets to systems: when it is time to build a database

Spreadsheets are an excellent tool and a terrible system of record. Knowing which one you are using is the whole question.

Six signs you have outgrown it

1. More than one person edits it. The moment two people need it at once, you have versions, and versions disagree.
2. It has a tab nobody understands. Usually built by someone who has left, containing a formula everything else depends on.
3. You cannot answer a question without opening several files. "How much of this did we sell last month" should take seconds.
4. Someone has typed the same customer four times, three ways. Without enforced uniqueness, the same entity multiplies, and every report becomes an argument.
5. There is no history. You cannot see who changed a figure, when, or what it was before. For anything financial or regulated, this alone is decisive.
6. It breaks when it grows. Large files get slow, formulas silently stop covering new rows, and errors propagate quietly.

Two or three of these is normal. Four or more and the spreadsheet is costing more than it saves.

What actually replaces it

Not "a database" on its own — that is a storage decision, not a solution. What replaces a working spreadsheet is three things:

• A data model where each thing exists once, with the rules enforced by the database rather than by hope. Two customers cannot share an ID; an order cannot exist without a customer.
• Screens for the jobs people do. Not a table view of the database. The three or four tasks that take most of the day, each one made fast.
• The reports people already use, rebuilt on live data so nobody exports anything to answer a question.

Skipping the third is the classic failure. People do not resist the new system because it is unfamiliar; they resist it because the report they relied on vanished.

Migrating without stopping

Run both for a while. Pick one process — the most painful one that is still small enough to be safe — and move it. For a fortnight, data goes into both and the outputs are compared. When the numbers agree, the spreadsheet for that process is retired, and you move to the next.

It feels slower than a cutover weekend. It is slower. It also does not fail catastrophically on a Monday morning with no way back, which is the usual alternative.

Expect the data to be messy

It always is. Duplicates, inconsistent spellings, dates in three formats, numbers stored as text. Cleaning is part of the work, not a surprise.

Insist on a reconciliation report: what was merged, what was corrected, what could not be matched and needs a human decision. A migration that silently decides things is a migration you will not trust, and trust is the entire point.

What it is not

It is not an ERP. Full enterprise systems take a year, cost a fortune and are abandoned more often than anyone admits. For most businesses the right first step is one well-modelled system covering one process properly, with room to grow.

If this sounds like your week, systems and databases is the service, and modernisation is how we move you without stopping the business.