Every business that’s been running long enough eventually reaches the same point: the system it depends on every day was built years ago, by people who may not even be around anymore, and it has quietly become the one thing nobody wants to touch. Not because it works badly. Often because it works, most of the time, and everyone is afraid of what happens the day it doesn’t.
We’re currently a few months into exactly that situation with a multi-store South African retailer. Their existing system has run purchasing and stock for years. It’s old, the technology behind it is no longer easy to find people for, and every change to it carries more risk than it should. It needed to be replaced. The harder question was how.
Why we didn’t do it all at once
The obvious approach to replacing an old system is to build the whole new one, test it, and switch over on a chosen date. It’s also, in our experience, the riskiest way to do it. A business like this one runs on its purchasing system every single day. If a full rewrite goes live and something in it doesn’t match how the business actually works, there’s no old system left to fall back on while it gets fixed. Everyone finds out at once, on the day it matters most.
So instead, we’re building the replacement, which we call iStoko, one working piece at a time, alongside the old system rather than instead of it. The old system keeps running production, untouched, while each new piece is built, tested against real data, and used by the actual people who’ll depend on it, before we move on to the next piece.
What “one piece at a time” actually looks like
We started with order capture, the part of the business where buyers place purchase orders with suppliers, because it’s a self-contained piece of work with a clear, testable outcome: can someone place a real order, correctly, faster than they could before. Once that was solid and signed off, we moved to the next piece. Each one gets tested against the same real data the business already has, not sample data invented for the demo, and each one gets used by the people who’ll actually be using it, not just approved by whoever commissioned it.
It’s slower to describe in a project plan than “go live in Q3”. It also means that at every point along the way, the business has a working system, not a promise of one.
It’s not really about this one retailer
The problem underneath all of this isn’t specific to one company or one industry. Any business that carries real stock, whether that’s a retailer, a wholesaler, or a manufacturer, eventually ends up with a system that started as the right answer and became the thing everyone is afraid to change. The shape of the problem, and the way out of it, tends to look the same wherever it turns up.
We’ll keep writing about how this project actually goes, including the parts that don’t go smoothly. Next up: what order capture involves once you get past the obvious bits.
Leave a Reply
You must be logged in to post a comment.