The custom software nobody maintains anymore

Plenty of companies run on custom software written ten or fifteen years ago. A capable employee built it, or a nephew who was good with computers, or a small local firm. Invoices, stock, customers — all of it goes through that program, every day.

It works. That’s the good part, and it’s also why nobody looks at it anymore. In the meantime the person who wrote it has left, moved away or closed their business. The software stayed.

As long as it works, nothing is urgent. But a few questions are worth answering before it stops:

1. Where is the source code? What’s installed on your computers is the finished program. The code it was built from is something else, and without it nothing can be changed. Often it’s still on the author’s laptop, out of your reach.

2. Which computer does it run on, and what happens if that computer dies? Usually it’s one particular machine in the office, running an old version of Windows that nobody dares to update. If it gives out tomorrow, the first thing to find out isn’t what a new one costs, but whether the program can still be installed on another.

3. Who has the passwords? For the server, the database, the domain, the hosting account. Sometimes they’re on a sheet of paper in a drawer. Sometimes they sit in a former employee’s inbox, which is also where the hosting invoices still go.

4. Can you get your data out? The customers, the invoice history, the stock — they belong to the company. But if the only place you can see them is the program’s own screen, they’re effectively locked inside. Worth trying once: can they be exported into a spreadsheet someone else could open?

5. What does the program actually do? Over ten years it has picked up rules nobody thinks about anymore: how a long-standing customer’s discount is worked out, what happens to a return, which number the next invoice gets. They aren’t written down anywhere else. You only notice them once they’re gone.

6. Who makes the next change? A VAT rate changes, a new reporting requirement arrives, a large customer wants invoices in a different format. The software has to change, and you don’t get to pick the deadline. If you don’t know who to call, that’s when you’ll find out.

What it means

Software doesn’t wear out like a machine. The code stays exactly as it was written. What changes is everything around it: the computer, the operating system, the law, the people who knew how it worked.

The risk grows quietly. Every trouble-free year makes it less likely that anyone will look at the program, and more expensive when someone finally has to. That moment usually comes with a deadline: a new regulation, a failed disk, a month to close.

The obvious fix isn’t safe either. A new program written from scratch loses exactly the rules from point 5 — the ones nobody remembers anymore.

What to do first

Almost never a rewrite. First the software is made safe, so it no longer depends on a single computer and a single person.

In practice, that means:

• the source code is recovered, as far as it still can be, and kept somewhere the company owns, not on one person’s laptop;
• the program is installed again, on a clean machine — the most honest test of the state it’s in;
• passwords and accounts move into the company’s name, with at least two people able to get in;
• the data is exported in a format any program can read, and the export is repeated regularly;
• the rules the program applies are written down, in plain words, in a company document;
• only then is anything decided: leave it as it is, keep maintaining it, or replace it piece by piece.

It isn’t a big project. At the end of it, the company knows exactly what it has.

Sometimes the conclusion is that the software is fine and should be left alone, with a copy of the data and someone who knows it. That’s a good answer, and far better found out now than on the morning the office computer won’t start. If you’d like to know where yours stands, get in touch.