When to go custom in your business: 5 signals that don't lie

I regularly get managers telling me "I don't know if we're ready for custom — we still have our tools". The answer is almost always: "You've been ready for 18 months, you just didn't dare to put it into words."

Here are the five signals that, in my experience, indicate it's time to leave the all-SaaS world to build (or automate) your tool. If three of them ring true, the answer is clear.

Signal 1 — You pay for what you don't use

The most common one. You subscribed to an ERP, a CRM or a SaaS platform 3-5 years ago. The tool does 80 things. You use 12. You pay for the 80.

Quick test:

Open your last recurring software invoice.
Next to it, list the 5 functions you actually use.
Divide the annual cost by 5.

If you get a number > €1500/function/year, you're overpaying. A custom tool that would only do these 5 functions typically pays back in 18 to 24 months — and belongs to you at the end.

Signal 2 — Your team builds workarounds to bypass the tool

This is the silent signal you never notice as a manager, because the team doesn't want to complain. But during audits, I see it everywhere:

A weekly Excel export to compensate for a missing view.
A Google Sheet shared alongside the CRM "because it's faster".
A file of mental post-its to "not forget" what the tool doesn't handle.

Each workaround is an invisible cost. Combined, they typically represent 3 to 10 hours per week per person — and as many possible errors.

Signal 3 — Your business process has become specific

When you signed with your SaaS, your business fit standard categories. Today, you've developed:

Custom pricing by client segment.
A production workflow that doesn't look like any sector's.
Compliance rules specific to your activity.
An invoicing logic your accountant has to reprocess by hand.

At this stage, you're no longer your SaaS's typical client. It slows you down more than it helps. This is exactly the moment when custom becomes profitable — because every specificity you've built is a competitive asset to protect.

Signal 4 — You just killed a SaaS project after 6 months

This one I often see in hindsight. The manager tested a new SaaS tool to solve a specific problem, realised after 4-6 months that the tool didn't fit, stopped, and lost both the implementation cost and the time spent.

If this has happened to you twice, stop the roulette. The cost of a properly scoped custom application (5 to 12 weeks) is often lower than two failed SaaS attempts — and you end up with a tool that does exactly what you need.

Signal 5 — You're dependent on a vendor

Three real cases:

The vendor raised prices by 40% in 2 years, "because they could".
The vendor was acquired, and the new team changed the product.
The vendor closed its "legacy" features — and yours were part of them.

When your business depends on software you don't control, you don't have a company — you have a tenancy. Custom isn't more expensive in the long run. It just makes you the owner.

Signals that do not mean you should go custom

To be honest, here are the cases where I advise against it:

You're under 5 people and your process isn't stabilised → stay on SaaS until your business is clear.
You're trying to cut a software cost of €2000/year → custom doesn't make economic sense at that scale.
Your problem is a missing feature, not the process → ask the vendor, not a developer.
Your team has no digital appetite → a custom tool won't save a team that doesn't want to use it.

The real cost: 2026 orders of magnitude

To give a concrete idea:

Project type	Duration	Typical budget
Targeted automation (1 process)	2-4 weeks	€1,500 - €4,000
Full internal tool (mini-CRM, client portal)	6-10 weeks	€5,000 - €12,000
Business application with integrations	10-16 weeks	€10,000 - €25,000
Light custom ERP	3-6 months	€20,000 - €60,000

To compare with: €200 to €2,500/month of SaaS over 3 to 5 years. In 70% of cases I see, custom pays back between 14 and 30 months.

The final test

Three questions:

Is your current tool slowing your growth? (not just "is it annoying")
Are you paying for workarounds — human or software — to keep it standing?
Would you sleep better if the code belonged to you?

Two yeses out of three = it's time to talk. Thirty minutes are enough to look at your case and honestly calculate whether custom is profitable — or not.