Off-the-Shelf vs Custom Software: An Honest Decision Guide

We build custom software for a living, so you'd expect this article to end with "build custom." It doesn't. Most of the businesses that ask us for a custom system would be better off buying something — and the ones who genuinely need a build usually already know it, they just want someone to confirm they're not being unreasonable.

Here's the honest version of the decision, including the cases where we tell people not to hire us.

When off-the-shelf is clearly the right call

Four conditions. If three or four of these are true, buy the product and don't look back.

• The process is standard, not strategic. Nobody wins customers by having a more interesting way of lodging a BAS. If a process is the same at your business as at every competitor, it's a solved problem — buy the solution.
• A mature product already fits about 80% of how you work. The remaining 20% is usually cheaper to change your process around than to build software for.
• Your team is small. Per-user pricing is genuinely cheap under about five people. At that size the maths almost never favours a build.
• You need it this month. A subscription starts on Tuesday. A build starts in six to twelve weeks.

Xero, payroll systems, email, document storage, e-signature, help desks: buy them. We use them ourselves, and when a client asks us to rebuild one of these we say no.

When custom is worth building

Five signals. One on its own isn't enough. Three or more and the sums usually work.

1. The process is your edgeThe way you quote, schedule, cost or inspect is the thing customers actually pay for, and every product forces you to do it their way instead.
2. Staff key the same data into two or three systemsRe-entry is the single most reliable indicator we see. It's pure cost, it introduces errors, and it never gets better on its own.
3. Spreadsheets are the real system of recordYou bought the software, and the business still runs out of a workbook that one person maintains and everyone is quietly frightened of. That workbook is a specification for the system you should have.
4. Licence fees are climbing faster than headcountPer-user, per-module, per-integration. At 15–30 users paying $60–$120 a month each, you're spending $15,000–$40,000 a year to rent something that still doesn't fit.
5. Your systems don't talk to each otherSometimes this is an integration problem rather than a rebuild — which is cheaper, and worth checking first.

The maths people skip

The buy-vs-build comparison is usually run as licence cost versus build cost. That's the wrong comparison, because it ignores the largest number in the equation: the cost of the workarounds.

Count them properly. Take each workaround — the re-keying, the manual reconciliation, the chasing, the "just check with Sarah" — and multiply the minutes by how often it happens by the loaded hourly cost of the person doing it.

Now put the three numbers side by side over five years: subscription fees, workaround labour, and a one-off build plus hosting and maintenance. For a team under five, off-the-shelf still wins. Between five and fifteen it's genuinely close and depends on how bad the workarounds are. Above fifteen, with real workarounds, custom usually wins by a wide margin — see our breakdown of what custom software actually costs in Australia for the build-side numbers.

The four honest failure modes

The answer is usually "both"

The version that works for most Australian SMEs isn't buy or build — it's a small custom core surrounded by bought products.

Keep Xero. Keep your email, your files, your payroll. Then build custom for the operational middle that no product handles: the jobs, the quoting logic, the scheduling, the field capture, the reporting your accountant's software will never produce. Then connect them so an approved quote becomes a Xero invoice without anyone typing it twice.

That's a smaller build than replacing everything, it's cheaper, it goes live sooner, and it leaves you on mature software for the boring parts where mature software is exactly what you want.

Five questions that settle it

1. Is this process the same at every competitor, or is it how we win?
2. Have we actually trialled the two best products, or just assumed they won't fit?
3. What are the workarounds costing us per year, in hours multiplied by wage?
4. Would an integration between what we already own solve 70% of this?
5. If we build, what is the smallest first version that would still be useful?

If the answers point to buying, buy — and we'll tell you which product to look at. If they point to building, build small and build the part that hurts.

Common questions