Build or buy: when custom software pays off

Buying is the right default. Building wins in a few specific situations, and it helps to recognise them before the licence renewal.

By the Maple Crew team · · 4 min read

We build custom software for a living, and our first advice to most clients is still to buy. Packaged software is cheaper, available today and maintained by someone else. Custom development has to earn its place. It does so in a small number of recognisable situations.

When the process is your advantage

If the way you do something is the reason clients choose you, forcing it into a generic tool erases the difference. Accounting is rarely a competitive advantage. The way you match candidates, assess risk or schedule field teams might be.

When the workarounds have become the system

Look for the spreadsheet that sits between two applications, updated by hand every morning. Look for the export, re-format, import routine that only one person understands. These are signs that the packaged tools no longer fit, and that you are already paying for custom software, in salaries.

When integration is the product

Often no single system is wrong. They just do not talk. A small, well-built integration layer between the tools you already own can deliver more than replacing any of them.

When licences scale worse than code

Per-seat pricing is attractive at twenty users and painful at two thousand. Run the five-year numbers, including the features you pay for and never use.

If you build, build small

  • Start with the one workflow that hurts most, not a platform.
  • Put working software in front of users within weeks and adjust.
  • Insist on tests, documentation and a deployment you could hand to another team.
  • Own the code and the data. Lock-in to a developer is still lock-in.

The honest answer to build or buy is usually both: buy the commodity, build the difference, and connect the two cleanly.

More insights

· 5 min read

Why your AI should not be the calculator

Language models write well and count badly. In credit analysis, the fix is architectural: code computes, the model explains.

Read article

Let's build something your team will use every day.

Tell us about the process you want to improve. We will reply with a first view of what is possible and what it would take.