When custom software is the wrong answer

The question nobody asks first

When a company approaches us about building an internal tool, the first thing we ask is whether they have genuinely evaluated buying one. Often the answer is that they looked at a product two years ago, or that a competitor mentioned something, or that a colleague thought the licence cost looked high.

That is not an evaluation, and getting it wrong is expensive in one direction far more than the other.

Buy when the process is not your advantage

Accounting, payroll, email marketing, help desk, HR records, basic CRM. These are solved problems with mature products, and your version of the process is almost certainly not a competitive advantage. Buying gets you a product improved continuously by a team larger than the one you could hire, for less than a month of custom development.

The licence cost is rarely the real comparison. Custom software costs what it costs to build, and then roughly fifteen to twenty percent of that annually, forever, to keep working.

Build when the process is the business

Custom becomes the right answer when one of these is true:

• The process is your differentiator. If how you schedule, price or fulfil is why customers choose you, forcing it into a generic product erodes the advantage.
• Integration is the actual product. Four systems that must behave as one, where the connective logic is where the value sits.
• The off-the-shelf option requires more workaround than use. If adopting a product means running two spreadsheets alongside it, you have not bought a solution.
• Per-seat pricing has stopped scaling. At sufficient headcount, licence cost genuinely can exceed build and maintenance. Model it honestly over five years rather than one.

The middle path most people miss

The usual right answer is neither: buy the commodity products and build only the thin layer that is genuinely yours. A well-integrated stack of three bought products plus one small custom service is cheaper to build, cheaper to run and far less risky than one large bespoke platform.

That is also the answer that earns us the least money, which is a reasonable indication it is honest.