What a custom ERP costs (and when a subscription wins instead)

If you're looking up what a custom ERP costs, you probably already have a subscription quote in front of you and want to know whether the alternative makes sense. The short answer is below, with numbers. But the part that matters is different: almost everyone runs the wrong comparison.

The comparison almost everyone gets wrong

A standard ERP quote shows a monthly subscription. A custom one shows a build cost. Those two numbers can't go on the same line: one is rent, the other is a purchase.

Put that way, the subscription always wins — in year one. The useful question isn't which number is lower today, but after how many years the two cross, and what happens to both curves in the meantime.

Because they don't rise the same way. A subscription grows with users and modules: hire three people and the bill goes up without anyone making a decision. Software you own doesn't grow with the number of people using it — it only grows when you decide to add something.

The numbers, as orders of magnitude

Nobody serious hands you a price list, because the price depends on how much you build. But an order of magnitude is fair game, and this is what I use when someone asks me on a call:

• A first working version — the piece you use every day, in production, with real data in it: roughly €15,000-40,000.
• A system covering operations — several departments, integrations with what you keep, roles and permissions: €40,000-100,000.
• Beyond that you build in phases, and anyone proposing a single monolithic hundred-thousand-euro project is selling you risk, not software.

On top of that comes annual maintenance, a percentage of the build cost. It isn't a hidden cost: it's the difference between living software and something nobody wants to touch in three years. I go into it in what business software costs to build.

The break-even point, in practice

Here's the calculation I suggest you run before any call — yours, without me. Take your current subscription, or the one you were quoted, and multiply by twelve. Then add, honestly:

• the modules you pay for that nobody opens;
• the customisations billed every time as if it were the first;
• the cost of the people retyping data by hand every month because the system doesn't do that thing;
• the increases from the last three renewals, projected onto the next three.

That number times five years is the real benchmark. In most cases I see, break-even lands between year three and year five — and from there the gap stops being a tie and becomes an advantage that widens every year.

Watch the last item on that list, because it's the one nobody books: if one person spends half a day a week copy-pasting between two systems, that's a recurring annual cost that appears on no quote.

When the subscription wins — and I'll say so

Custom isn't always the right call, and anyone who sells it to you every time isn't looking at your numbers.

• If your process is everyone's process.Invoices, stock, deadlines, bookkeeping: if you work the way everyone works, an off-the-shelf product costs less and does the job. Buy that.
• If you're few and staying few. The custom advantage compounds with users. Below a certain size, break-even arrives too late to matter.
• If the process changes every six months.Custom software photographs a way of working. If that way isn't stable yet, you crystallise the chaos instead of fixing it.
• If the horizon is short.A sale, a merger, a change of ownership on the way: a five-year break-even isn't your problem.

What actually blows up the quote

In my experience quotes don't blow up because of the number of screens. They blow up for three reasons, and they're worth asking about before you sign.

• Integrations.Making the system talk to e-invoicing, e-commerce or production machines depends on how the other side is built — and neither of you controls that. It's the line item that varies most.
• Migrating the history.Twenty years of data don't move with an import. They have to be understood, cleaned, verified. Anyone quoting it as a single line hasn't looked at it yet.
• The scope nobody wrote down."And then obviously there'll also be…": everything after thatobviously is what turns a quote into an argument.

Before you ask for a quote

The cheapest way not to get this wrong isn't getting three quotes to compare — the advice everyone gives, and nearly useless, because without technical skills three quotes aren't comparable. It's starting from a technical audit: two or three days, a written report that's yours either way, and from there scope gets defined with numbers under it instead of by feel.

If you want to see how the actual work goes — migration without stopping the company, code you own, no per-seat licences — I cover it in custom ERP. If instead the piece that chafes is the sales side, the reasoning is the same and you'll find it on the custom CRM.