SaaS vs custom software: the SME calculation that changes everything
SaaS is a good deal when you truly use the product.
It becomes much less impressive when you pay 400, 600 or 1,200 euros a month to use three features out of two hundred, then your team completes the rest in Excel.
That is where many SMEs get the calculation wrong. They compare “monthly subscription” with “custom development”. They forget the third cost: the human time spent adapting the company to a tool that was not designed for it.
The real SaaS vs custom software debate is not ideological. It is not “SaaS is bad” or “custom is always better”. The useful question is simpler: does the tool fit your process, or does your team work around the tool?
The 14,400 euro per year SaaS used like Excel
I saw a very clear case: a client was paying 14,400 euros a year for a CRM.
When we looked at the real usage, the team did three things: entering contacts, logging meetings and exporting data to Excel for reporting.
That was it.
They were using Salesforce like a spreadsheet with a login.
14,400 euros per year. For several years. With human time on top to reshape exports, correct data and work around what the tool did not model properly.
Salesforce was not the problem. Salesforce is a strong product for organisations that use its power. The problem was the gap between the product being paid for and the actual business need.
A custom app solved the three useful actions: contacts, meetings, stats. Nothing more. But exactly in the way the client worked.
Cost of the first brick: 5,000 euros before tax. Once.
When SaaS is the right choice
SaaS is still often the best choice.
If your need is standard, if the product covers 80% of your process, if integrations are clean, if your team adopts the interface easily, use SaaS.
Accounting, payroll, email marketing, customer support, booking: many categories are already very well served. Building custom software to recreate a mature product worse would be a bad decision.
No-code and SaaS tools can also be enough for a simple prototype, a landing page, a form, a quick test or a temporary workflow.
5000.dev is not anti-tool. Custom software only makes sense when it removes more friction than it creates.
When custom becomes rational
The warning sign appears when your team starts bypassing the SaaS.
You export to Excel every week. You add manual columns. You keep a side procedure because the tool does not match reality. You pay for licences for features nobody opens. You need one precise action, but the vendor pushes you toward a higher plan or a heavy integration.
At that point, you are no longer just paying for software. You are paying for software plus the human work required to compensate for what it does not do.
It is the same pattern we see with SME business apps: information exists somewhere, gets re-entered elsewhere, then disappears into an email or spreadsheet.
Custom becomes rational when the process is specific, recurring, stable and directly tied to margin, sales speed or service quality.
You do not need to create a whole platform. Often, you need a first brick:
- one screen centralising the useful data;
- one business action replacing five manual steps;
- a tracker that matches the way you sell, produce or deliver;
- an export in exactly the right format;
- a simple workspace for your team or clients.
It is rarely spectacular. It is often very profitable.
AI changes the make-or-buy calculation
For a long time, custom software was reserved for heavy budgets. Six months of specs, eighteen months of development, five years of maintenance. Many SMEs chose SaaS by default. That made sense.
AI changed the economics of code.
Not because a non-developer can safely build anything alone through vibe coding. That becomes risky when data, permissions, payments or security enter the picture. AI mainly changes the productivity of a senior developer who knows how to scope, cut and ship.
What was too expensive yesterday can become reasonable today if the scope is split correctly.
At 5000.dev, the decision unit is intentionally simple: one custom web app brick, 5,000 euros before tax, 2 weeks of development, delivered or refunded. If the need is larger, we do not sell a 6-month tunnel. We split it.
That is where the SaaS vs custom calculation becomes concrete.
A SaaS at 600 euros per month costs 7,200 euros per year. If it really solves the problem, keep it. If it forces your team to work around it, a 5,000 euro custom brick can become more rational in the first year.
The SME decision grid
Before choosing, ask the questions in this order.
Is your need standard? If yes, look for SaaS first.
Do you use at least 70 to 80% of the product? If yes, SaaS probably still makes sense.
Does your team often export data to work elsewhere? If yes, the tool does not really cover the process.
Does the human work around the SaaS cost more than the subscription? If yes, the displayed price is misleading.
Is the process stable enough to encode? If everything changes every week, wait. If the same action is repeated every day, consider custom.
Can the first version fit into one brick? If yes, you do not need a large project. You need one useful first app.
For spreadsheet-heavy workflows, the article replace Excel with an app explains the operational calculation.
FAQ
Is custom software always more expensive than SaaS?
No. A poorly used SaaS can cost more over a year than a first custom brick, especially once you include the human time spent working around it.
When should I keep SaaS?
When it covers most of the need, the team uses it well and it avoids maintaining a specific system. Custom software should not be automatic.
Is building our own business software risky?
The risk mostly comes from oversized projects. A first brick limited to 5,000 euros before tax and 2 weeks keeps the commitment small and lets you validate quickly.
If your SaaS increasingly feels like a paid subscription used to feed spreadsheets, the useful move is to calculate the lost time and compare it with one genuinely useful custom brick.