4 min read
Buy a system off the shelf, or build one?
Six criteria for choosing between a tool that already exists and one built for you, plus a simple rule for knowing which side you're on.
It's the first decision and almost nobody makes it with a clear head. People choose based on what a friend did, or whatever the first salesperson pitched them.
And it's worth approaching with healthy fear. The Standish Group's CHAOS report (opens in a new tab), the field's reference measurement, found that software projects that ship but run into trouble average 189% over the original cost. Nearly triple. Getting this choice right, before you start, avoids most of that risk, because the cheapest project is the one you don't do.
The short rule: if what you do looks like what thousands of other companies do, buy something that already exists. If your edge lives exactly there, build it.
Everything else is applying that with more detail. Six criteria, in order of importance.
1. How much of what you need is standard?
Invoicing, collecting payments, selling online, booking appointments, running payroll. Thousands of companies do it the same way you do, dozens of tools compete to solve it, and none of them will charge you what it costs to build it yourself.
If your list of needs is eighty percent standard and twenty percent particular to you, the answer is almost never to build everything custom. It's to buy the standard part and build only that twenty percent, connected to the rest.
2. Is your particular case actually particular?
This is the most expensive mistake, and it cuts both ways.
Many companies believe their way of working is unique when it's really just a habit. "We do it this way" doesn't always mean "we need it to work this way": sometimes it means nobody ever questioned it.
Before paying to reproduce a quirk inside a system, ask whether you'd defend it in front of a customer. If the honest answer is "it's just what we've always done," it's probably cheaper to adapt to the tool and skip the development.
3. How much will you fight the tool?
Every off-the-shelf tool imposes its own way of working. That's fine as long as the gap is small.
It gets expensive when the team starts inventing workarounds to dodge it: fields repurposed for something they weren't meant for, parallel spreadsheets, a notebook next to the screen. Every workaround is a sign the tool doesn't fit, and it usually costs more in wasted hours than fixing it properly would.
4. Whose data is it, and how do you get it out?
Before signing up for any tool, find out how you'd export your data if you ever wanted to leave. It's not distrust, it's the bare minimum.
If the answer is "you can't" or "you have to request it through support," treat that as a future cost. Switching tools with your data trapped inside is the kind of problem that can stall a company for months.
5. What does it cost you three years out?
An off-the-shelf tool is paid monthly, per user. A system built for you is paid mostly upfront, then maintenance.
Run the numbers over three years, not one. Sometimes the custom build, which looks expensive, ends up costing less than fifteen people's subscriptions over that stretch. And sometimes it's exactly the opposite. What matters is that it's a calculation, not a gut feeling.
6. How long can you wait?
An off-the-shelf tool works next week. A custom build takes months.
If the problem is urgent, buy something off the shelf now even if it isn't ideal, and decide calmly later. Getting the urgency under control with an imperfect solution beats waiting six months for the perfect one while the problem keeps costing you.
The option almost nobody considers
It isn't black or white. The combination that works best for most companies is buying the standard part off the shelf and building only the piece that sets you apart, with the two connected.
You pay for development only where it gives you an edge, and let someone else maintain the rest. It's cheaper, faster, and easier to change later.
A quick way to decide
Write down everything you need the system to do, one line each. Mark every line S for standard or P for particular to you.
If it's almost all S, buy. If it's almost all P, build. If it's mixed, which is the most common case, you now know exactly what to buy and what to build.