Should my business build custom software or keep paying for SaaS subscriptions?

The short answer: keep your SaaS subscriptions for anything commoditised, accounting, payroll, email, video calls, because those are solved problems and no custom build will beat Xero at being Xero. Consider building custom for the workflows that are specific to how your business operates, especially where you are currently paying for several subscriptions that each do part of the job while a person manually bridges the gaps between them. That middle ground, the connective work between your systems, is where the economics have shifted dramatically in the last two years, and it is where most of the businesses we work with get their best return.

The traditional advice on this question was simple: small businesses buy, big businesses build, because custom software used to cost six figures and take a year. That advice is now out of date, and the reason is that AI has collapsed both the cost and the timeline of custom development. A build that would have cost $150,000 three years ago can now be delivered for a fraction of that, which changes the maths entirely for smaller organisations. I wrote separately about what this shift means for the SaaS industry itself, but the practical version for a business owner is that the build option is now on the table at sizes where it never used to be.

A framework for the decision

First, ask whether the function is generic or specific. Payroll is generic, and your way of quoting jobs, onboarding contractors, or processing enrolments is probably specific. Generic functions belong in SaaS. Second, ask what the real cost of your current setup is, which is not just the subscription fees but the hours someone spends every week moving information between tools that do not talk to each other. For a lot of businesses, that hidden labour cost is several times the subscription cost. Third, ask whether the way you do this work is part of why clients choose you. If your process is your edge, renting a generic version of it from a vendor means operating at the same ceiling as everyone else who rents it.

The subscription maths for a small business

The subscription maths matters more for small businesses than large ones. If you are paying $200 per user per month across four tools for a team of ten, that is $96,000 over four years, for software that was never quite right and where a person still fills the gaps by hand. A custom build in the $8,000 to $25,000 range that replaces two of those tools and eliminates the manual bridging work often pays for itself inside the first year, and after that the comparison is not close.

When you should not build

Where I would caution against building: if a well-known tool does 80 percent of what you need and the missing 20 percent is an annoyance rather than a cost, keep the tool. If your process is still changing month to month because the business is young, build later, once the process has settled. And never build a custom version of something where the SaaS product's ongoing development is the value, because you do not want to be maintaining your own accounting compliance updates.

Questions we get asked

Do we have to replace our existing systems to go custom? No, and usually you should not. The best-value builds sit between your existing systems and connect them rather than replacing anything.

What about maintenance costs? Budget 15 to 20 percent of the build cost annually at the high end. For a focused workflow build it is usually less, and it is predictable rather than per-seat.

Is it risky relying on a small development partner rather than a big vendor? Ask any prospective partner two things: who owns the code (you should) and what documentation is handed over. If the answers are you and everything, the risk is manageable and lower than being locked into a vendor's roadmap.