We’ll try to talk you out of this first. Custom software is the most expensive option, it carries the longest tail of ownership cost, and a good three-quarters of the time the outcome is reachable by configuring and connecting things you already pay for.
But sometimes it genuinely isn’t. Your eligibility rules are unlike anyone else’s. The workflow is your actual competitive advantage. Every product evaluated needs so much bending that you’re maintaining a fragile customisation anyway. In those cases building something small and precise is the cheaper path.
How we approach it:
- Scope down hard. The first release covers the process that hurts, not the roadmap. Most failed internal apps failed by being too ambitious in version one.
- Boring, well-supported technology that another developer can pick up. Nothing clever, nothing exotic.
- Integrated, not isolated. It reads and writes to your existing systems so it doesn’t become another data island.
- Built to be handed over — documented, with access and source in your name from day one, not ours.
- Planned for year two. Who maintains it, what it costs to run, and how changes get made after we’ve gone.
If part of the requirement can be met with a low-code build or an off-the-shelf tool, we’ll split it that way and only custom-build the part that has to be. Smaller custom surface area is almost always the right answer.