The spreadsheet is the system

Hank has been telling me for a year to write these down, on the grounds that I keep explaining the same four things and would save myself the breath. So, the first one, which is the thing I see in nearly every company I walk into.

It lives on a shared drive. It has a name like pricing_v4_final. There are two people who know not to sort column D, and one of them is on holiday. Nobody chose this. It grew the way paths grow across a lawn, because it was the shortest way to get something done on a Tuesday, and then it was still there on Friday, and then it was February.

A spreadsheet has become load-bearing when three things are true: more than one person depends on the output, the logic inside it is recorded nowhere else, and a mistake in it would reach a customer. At that point it is not a document. It is an application with no tests, no backup and a single maintainer who does not know that is her job.

The instinct is to replace it immediately, and I have learned to talk people out of that. It holds years of decisions nobody wrote down, and rebuilt in a hurry it will cost ten times the price and be worse. Do the cheap things first. Put it somewhere with version history, lock the formula cells, write one page explaining the strange columns, name an owner and a deputy. That is an afternoon, and it removes most of the risk. Then you can decide calmly, and not in a panic, whether it deserves to become software.

MMB