Every operations leader I've worked with eventually hits the same wall. The SaaS tool that was perfect for the company at 20 staff is in the way at 80. They patch around it with Zapier and shared spreadsheets. By 150 staff they're running on duct tape, and the question becomes: do we keep buying, or do we build?
The wrong answer in either direction costs six figures. Here's a working framework.
When SaaS wins
SaaS is the right answer when three things are true:
- Your workflow looks like everyone else's. If your invoicing, accounting, or HR process matches what a thousand other Australian businesses do, somebody has already built better software for it than you can afford to.
- You'd rather rent than maintain. No engineering team, no on-call, no security updates, no migrations.
- The total cost of ownership is below the build threshold. A $200/month SaaS over five years is $12,000. A $50,000 custom build only beats that if it lasts five years and saves real time on top.
For 80% of standard business processes, accounting, payroll, CRM, helpdesk, HRIS, SaaS will win every time. Building Xero from scratch is not a productive use of your engineering budget.
When custom wins
The decision flips when:
- Your workflow is unusual. If you make money in a way no other company does, you don't have a SaaS competitor. You have a Frankenstein stack of three SaaS tools and four spreadsheets pretending to be one system.
- The SaaS roadmap doesn't match yours. You've requested the same feature for two years, the SaaS company says "great idea, on the roadmap," and it never ships. Custom doesn't have a roadmap to disagree with.
- Per-user pricing is killing you. Above a few hundred staff, per-user SaaS often crosses the threshold where a custom build pays back inside three years.
- You need integrations the SaaS doesn't provide. If you're paying $400/month for Zapier to glue four tools together, you've already started building custom; you just haven't admitted it yet.
SaaS is rent. Custom is mortgage. Both are valid. The question is whether your business is going to live in the building long enough to justify owning it.
The hybrid that usually wins
The best operating model I see in growing Australian businesses isn't pure SaaS or pure custom. It's:
- Buy commodity: accounting, payroll, email, helpdesk, marketing automation, all SaaS.
- Build the unique 10–20%: the workflow that makes your business your business.
- Glue them together: proper integrations, not Zapier chains, when reliability matters.
The custom layer is small but it's where your competitive advantage lives. Don't outsource it; don't try to find a SaaS for it.
Questions to ask before you build
- How many SaaS products have you tried that didn't fit? If the answer is fewer than three, keep buying.
- What specifically don't they do? Write it down. If you can't write down the gap, you don't have a custom build problem yet. You have a configuration problem.
- What's the workflow worth in revenue? If the workflow generates $X/year, the build budget should be a fraction of that, not a multiple.
- Who maintains it? If the answer is "we'll figure that out later," don't build.
Hidden costs of each path
SaaS hidden costs:
- Per-user pricing inflation as you scale.
- Integration glue (Zapier, Make, custom scripts) compounding monthly.
- Workarounds for missing features eating staff time.
- Vendor lock-in that prices a future migration at $50k+.
Custom hidden costs:
- Maintenance forever (15–25% of build cost per year is normal).
- Documentation drift. Whoever built it leaves. Nobody knows how it works.
- Slow feature evolution because the team is small and busy.
- Security and compliance work that SaaS hides from you.
The bottom line
Buy SaaS for everything that looks the same as everyone else's business. Build custom for the unique workflow that makes you money. Glue them together properly when the boundary moves.
If you want three quotes from Australian developers who'll be honest about whether you should build or just configure, FindDevs gets you that. Free, three real opinions, in 24 hours.