When should you choose custom-developed software?

Before we talk about when to build, it's worth saying when not to: in the majority of cases. CRM, accounting, project management, helpdesk, email, calendar, meeting booking, invoicing, document handling — if you're doing something that resembles what other companies do, there's usually a standard product that solves it 80–90% correctly.

The remaining 10–20% is rarely worth building for. Standard software has an entire industry's learning baked in: integrations that work, a community discussing the bugs before you do, updates that arrive without you having to order them, and a price already pressured by competition. Building your own to avoid a small compromise is one of the most expensive mistakes an SMB can make.

The burden of proof therefore sits on "build" — not on "buy". If you can't articulate clearly what you get from owning the code that you can't get any other way, the answer is usually that you shouldn't build.

The threshold between off-the-shelf and custom is rarely sudden. It creeps up. You start with a workaround. Then another. Then someone builds an integration around the tool. And suddenly three people spend 10–20% of their time keeping Notion, Airtable and a Zapier flow alive, because the original setup no longer fits.

That's the point — not before — when the conversation about custom makes sense. The threshold is reached when how you actually work materially diverges from what the standard tool was designed for, and when that divergence is a business strength you want to develop. If the divergence is merely annoying but doesn't give you a business advantage, building is still the wrong call.

We typically test the threshold with three questions: has the workaround we built become part of the core process? Is it slowing us down on the things we actually want to differentiate on? And — most importantly — is anyone in the organisation ready to own a custom solution for five years, not just commission it?

When people compare off-the-shelf and custom, they almost always compare licence cost to development budget. That's wrong. Custom has no licence — but it has operations, maintenance, updates, security patches, monitoring, debugging, change requests, and — most importantly — dependence on the people who know where the buttons are.

A good custom project assumes that the real ownership cost over five years is a multiple of the build cost. That's not scare tactics; it's just how software works. A platform that stays up for five years has shifted stack versions, added new integrations, weathered ten rounds of compliance change, and had its features revisited. All of that needs to be done by someone.

Off-the-shelf software removes that burden and parks it with the vendor. You pay licence instead — but you also pay in less control, slower change, and being one of many customers the vendor prioritises between. It's an honest trade; it's just worth making consciously.