Custom Web App vs Off-the-Shelf SaaS: When Is Custom Worth It?

Somewhere in your business there's a spreadsheet held together with hope, or a stack of SaaS subscriptions that almost fits how you work. At some point someone asks: should we just build our own tool?

I build custom web apps for a living - mine start at €2,000 - so you'd expect me to say yes. But the honest answer is that most businesses asking this question should buy SaaS, and a specific minority should absolutely build. The trick is knowing which one you are before you spend money. Here's the framework I actually use when someone brings me this question, including the cases where I talk them out of hiring me.

The decision framework

Four questions decide almost every case:

1. Is this workflow how you win, or just how you operate? Accounting, email, calendars, payroll - every business does these roughly the same way, and SaaS products for them are polished by thousands of customers' feedback. Buy. But the workflow that makes customers pick you - your quoting logic, your production scheduling, your service process - is exactly where forcing yourself into someone else's software sands off your edge. That's build territory.

2. How much are you bending your process to fit the tool? A little configuration is normal. But if your team maintains spreadsheet workarounds around the SaaS, exports data to manipulate it elsewhere, or has "the way the tool wants it" and "the way we actually do it" as two separate things - you're already paying custom-software costs in labour, without getting custom software.

3. What does the subscription stack actually total? Not one tool - the stack. A typical small business runs a CRM (€50-150/month), a scheduling tool (€30-80), forms (€30-50), a client portal (€50-100), and Zapier duct tape holding them together (€30-100). Per seat, growing with headcount. Individually reasonable, collectively often €400-800 a month for a workflow one focused app could handle.

4. How many people, how often? Custom makes sense for tools used daily by your team or customers. It rarely makes sense for something three people touch once a month.

The 3-year math

Sticker prices lie in both directions, so compare total cost over three years.

SaaS: €300/month across a modest stack for a 5-person team = €10,800 over three years. Plus per-seat growth as you hire, plus price increases (SaaS pricing only moves one way), plus the quiet labour cost of workarounds. Call it €12,000 to €20,000 for a stack that fits you 80 percent.

Custom: A focused internal tool from me starts at €2,000; a realistic mid-size app - accounts, dashboard, a couple of integrations - runs €3,000 to €8,000 from a good freelancer (agencies: double to quadruple). Running costs on a modern stack are genuinely low: €10 to €50 a month hosting, plus maintenance - budget €500 to €1,500 a year. Three-year total for a €5,000 build: roughly €7,000 to €10,000, for software that fits you 100 percent, with no per-seat meter.

The crossover is real but it's not universal. If your SaaS spend on a workflow is under ~€100/month and the tool fits well, SaaS wins the math almost every time. If you're at €300+/month and fighting the tools, custom usually wins by year two - and keeps winning every year after, because a build is a one-off and subscriptions are forever.

One honest caveat on the custom side: the first version won't be as polished as a SaaS product with a decade of iteration. It will do your workflow better on day one, but budget emotionally for a few rounds of refinement in the first months.

Integration and lock-in: the costs nobody quotes

SaaS lock-in is gradual and real. Your data accumulates in their format, your team's habits form around their UI, and export tools are always mysteriously worse than import tools. Leaving after three years is a migration project. And you're exposed to their roadmap: features you rely on get deprecated, the pricing tier you're on gets "sunset," or the product gets acquired and coasts. You have no vote.

Custom lock-in is a different shape: you're tied to whoever maintains it. This is the legitimate fear, and the mitigation is contractual and technical, not luck. You should own the code and the repositories outright, it should be built on boring mainstream technology (I build on Next.js, TypeScript and Postgres precisely because thousands of developers can pick it up), and it should come documented. A custom app you own on a mainstream stack is less locked-in than a SaaS subscription - any competent developer can take it over; no one can take over Salesforce's codebase for you.

Integrations favour custom more than people expect. SaaS tools integrate on their terms - whatever Zapier triggers they expose. A custom app integrates on yours: it talks to your accounting software, your inventory, your payment provider exactly the way your process needs. Half the custom apps I build exist primarily to be the glue their SaaS stack refused to be.

When SaaS honestly wins

So there's no ambiguity, here are the cases where I tell people not to hire me:

• Solved problems. Accounting, email marketing, generic project boards, HR. You will not out-build Xero with any budget worth spending.
• Unvalidated workflows. If you don't yet know how your process should work, a €30/month tool is a cheap way to find out. Build custom after the process stabilises, not before.
• Compliance-heavy domains. Payroll and tax software encode regulation that changes yearly. Let a company with a compliance team own that burden.
• You need it next week. SaaS is live this afternoon. Custom takes weeks.

I follow my own advice here: I run my own products - MyHomeStock, PDFer, QRify, DevOpsNess and others - on custom code because the product is the business, but my accounting is a subscription like everyone else's.

The short version

Buy SaaS for the workflows every business shares. Build custom when the workflow is your edge, when your subscription stack for it passes €300/month, or when you're visibly bending your process to fit tools that almost fit. Over three years, a €2,000 to €8,000 custom build frequently beats the stack it replaces on raw cost - and always beats it on fit - provided you own the code on a mainstream stack.

If you're weighing this decision, describe the workflow to me and I'll give you a straight answer with numbers - and if the honest answer is "just use a €40/month tool," that's what I'll tell you. How I build web apps is on the web apps page, and my fixed prices are on the pricing page.