Custom software vs off-the-shelf: how to decide.
Every growing business eventually asks whether to keep paying for a tool that doesn't quite fit, or build something of its own. Software companies will usually tell you to buy; development agencies will usually tell you to build. Here is a more honest way to decide.
By Usama bin Mazhar · 6 min read · Updated
Buy when the process is standard
If your process works the same way it does at thousands of other businesses — payroll, accounting, email marketing, standard helpdesk — buy. Someone has already spent years perfecting that software, and you benefit from every improvement they make.
Buying is also right when you need something working this month, when you have no one to own a custom system, or when the budget is small.
Build when the process is your advantage
Build when the way you work is part of why customers choose you. Forcing a distinctive process into generic software makes you more like your competitors, not less.
Building also makes sense when you are stitching together several tools with manual work in between, when per-user licence costs keep growing faster than the value, when you need to control where data lives, or when the software you need simply doesn't exist.
Compare cost over time, not the first invoice
Off-the-shelf looks cheaper because the cost is spread out. Custom looks expensive because it's paid up front. Compare them over three to five years, including:
- Licence fees as your team grows, and likely price increases
- Staff time spent on workarounds and duplicate data entry
- Integration and customisation costs for the bought tool
- Hosting, maintenance and improvement costs for the custom system
- The cost of switching later, in either direction
The option most people miss: both
The best answer is often a combination: buy proven software for the standard parts, and build a thin custom layer — an integration, a portal, an automation, a dashboard — for the part that makes you different. You get reliability where it's standard and fit where it matters, without building everything.
Questions worth answering before you ask them.
Up front, usually yes. Over several years, not necessarily — especially when licence fees scale with headcount or a poor fit costs hours of manual work every week.
This is a real risk to manage. Insist on owning the code, using mainstream technology, and getting documentation — so any competent developer can take over.