Build versus buy: when custom software is the cheaper option

The build-versus-buy conversation usually compares a development quote against a licence fee, concludes that buying is cheaper, and stops. The comparison is wrong in both directions — it ignores the real cost of adapting to a product, and it ignores the real cost of maintaining something you built.

Buy is right more often than engineers admit

• The process is genuinely standard — accounting, payroll, email, CRM basics, helpdesk. Your version will not be better and the product improves without you.
• Compliance and security are someone else's full-time job.
• You need it working next month, not next year.
• Nothing about how you do it is a competitive advantage.

Build is right more often than finance admits

• The process is how you actually win — the thing your customers notice and competitors cannot copy from a vendor's feature list.
• The off-the-shelf option needs so much configuration and custom development that you are building anyway, but on someone else's constraints.
• Per-seat pricing scales with your growth in a way that turns a small line item into a large one within three years.
• You need it to sit between systems that no vendor connects, which is where most integration-heavy custom work earns out.

The costs nobody quotes

On the buy side: implementation and data migration, integration work, per-seat growth, the annual uplift, training, and the cost of changing your process to fit the tool. On the build side: hosting, monitoring, security patching, the second developer who has to understand it, and the rewrite that arrives when the original framework goes out of support. Model both over three years, not one.

The middle path most companies should take

Buy the commodity, build the differentiator, and invest in the integration between them. In practice that means a standard CRM and standard accounting, with custom software for the workflow that is specific to your business, connected properly by an API layer you own. That combination is almost always cheaper over three years than either extreme, and it is the one that survives a change of vendor.