From a critical spreadsheet to an API: getting knowledge out of the sheets

In every technical company there is the spreadsheet: that one file which calculates something that bids, projects or purchasing decisions depend on. Someone competent built it years ago, it has survived three template versions, and today nobody touches certain cells. It works — and precisely for that reason nobody reviews what is inside it.

There is nothing wrong with knowledge starting out in a spreadsheet: it is an excellent tool for thinking with numbers. The problem is when the spreadsheet stops being a draft and becomes infrastructure without anyone having treated it as such.

The signs that yours has already become a risk

• There is only one person who “understands it”, and their holidays show up in your timelines.
• Several copies circulate with names like calc_v7_FINAL_good.xlsx, and nobody can say for certain which one was used in that bid back in March.
• A change in the standards or in a rate forces you to review formulas by hand, cell by cell, with no guarantee that none was missed.
• Results are copied into other documents by hand — and at some point they have been copied wrong.
• Nobody can answer “who calculated this, when, and with what data?” without a laborious search.

Three or more signs: the spreadsheet has stopped being a tool and has become a risk.

What turning it into a service means

The alternative is not “moving it to another program”, but changing its nature: taking the rules out of the cells and turning them into a service with three properties that a spreadsheet cannot provide.

• Validated inputs. The service rejects impossible data before calculating; you avoid seemingly reasonable results built on a typo.
• Versioning. Every change to a rule is recorded: you know which version calculated what, and a revision of the standards is a controlled update, not a complex migration.
• Traceability. Every run leaves a record of who ran it, when, and with what data, so the audit question is answered in seconds.

And a consequence that often goes unnoticed: once the calculation is a service, any other tool can use it. The document that used to receive the result by manual copying now requests it directly; transcription errors disappear by design. That is how the expert knowledge as an API case is built, with the technical memoranda generated from that same engine.

The path is incremental

There is no need to migrate everything. The order that works: first the most critical or most used calculation, validated against the spreadsheet itself — same input, same result, demonstrated with historical cases. The spreadsheet is not discarded: it stays as a reference and a backup while the service gains ground, one calculation at a time. Within a few weeks, what is critical is under control; what is secondary can wait its turn or stay where it is indefinitely. That first calculation is a narrowly scoped deliverable with a fixed price; if your company operates in industry or technical services, automation for industry and technical companies sets out the full path.

The spreadsheet did its job for years. Turning it into a service is not a matter of distrusting it, but of giving it the solidity its role already demands.