What a custom ERP actually costs in Romania

If you have ever searched for what an ERP costs, you already know the answer you get: "it depends". That is not a bad-faith dodge — it genuinely does depend. But there is a lot of room between "it depends" and silence, and this article tries to fill it: what the cost is actually made of, what moves it, and what you can work out on your own before you ask for a first quote.

Why nobody tells you the price

Search for the Romanian equivalent of "how much does an ERP cost" and look at what comes back. The large local vendors publish entire pages about why a price cannot be given. Senior Software devotes a guide to explaining at length why no figure can be quoted on the spot, and ends with an invitation to talk. Pluriva does the same. You will find the same pattern at almost every vendor: an article about the factors that influence cost, then a form.

Their reason is real. A wrongly sized ERP costs more than an expensive one: you pay for licences you never use, or you discover halfway through the rollout that the one module everything depended on is missing. A price given before anyone understands the workflow is a figure that will change anyway.

But the practical consequence for you, the person searching, is that you are left without a reference point. You cannot tell whether the conversation is about ten thousand euros or a hundred thousand, so you cannot even tell whether it is worth starting. What follows tries to give you that reference point, without selling you a fake number.

What the cost is actually made of

The price of a custom system is not a number, it is a sum of things that can be counted separately. When a quote arrives, these are the lines that should be visible in it.

Analysis. Someone has to sit with the people who do the work and understand how it is done today. Not what the procedure says — what actually happens, including the shortcuts everyone has invented for themselves. It is the part most often skipped, and the one that produces the most expensive surprises.

The modules. A ticketing system is one thing; tickets plus fleet plus invoicing plus timesheets is another. Every module has its own screens, its own rules and its own edge cases. This is where the figure grows most visibly.

The integrations. The tax authority, the bank, the payment processor, the couriers, an old system that has to keep running for a while. Each integration is a contract with a system you do not control: it has its own format, its own errors and its own pace of change. It costs more than it looks, because most of the work is testing what happens when the other system answers differently from what its documentation says.

Data migration. Data from the old system has to be read, cleaned and checked. If it comes out of spreadsheets kept by several people, cleaning it is a project of its own. This is the line quotes underestimate most often.

Training and go-live. A system nobody knows how to use has not been delivered. Someone has to show people how the work is done inside it, and stay close during the first weeks.

Running it afterwards. Hosting, verified backups, monitoring, updates. It can be bundled or billed separately — but it exists either way, so it is worth knowing which part of the quote it sits in.

What pushes the figure up, and what keeps it down

Up: the number of modules; integrations with systems that are not yours; the number of distinct roles, because each role means another set of screens and another set of rules about what it is allowed to see; old, dirty data from several sources; and the requirements that appear once you have seen the first version working — normal, but still new work.

Down: a narrow first stage that solves one thing and solves it completely; processes that are already settled, which people carry out the same way; clean data; and one person deciding, rather than a committee negotiating every screen.

The most effective thing you can do for the budget is not to negotiate the rate. It is to cut the requirements list in two: what it cannot work without, and what would be good to have. The first list gets built now. The second gets built after the first one runs and shows you what is genuinely missing — which is usually not what you thought.

What delivered projects show

I cannot tell you what your platform costs without knowing what you do. But I can say what it meant concretely, as a volume of work, in three systems that are running.

Worklio is an operational ERP for service and maintenance: tickets with a due date and an owner, a fleet with the day's route on the technician's phone, and invoicing that starts from the signed work order. Three modules that feed each other — the job written down once travels on its own all the way to the invoice.

Operations covers transport and construction: trips and kilometres, monthly timesheets, site modules with their own rate catalogue, and a financial panel that composes profit per truck out of data already entered, rather than out of a calculation done by hand at month end.

FlowDeskOne works the commercial side: a request becomes a quote, the quote becomes a contract, the contract becomes a work order and the work order becomes an invoice validated against the RO_CIUS standard and sent automatically into the tax authority's private virtual space. Plus stock, goods receipts and the e-Transport declaration.

Look at them not as a feature list, but as a way of measuring your own project: how many of those blocks you would need, and which of them is the one without which the rest makes no sense. That one is your first stage.

What you can do now

Write down on one page what does not work today and what it costs you — not in money, but in observable things: how many times the same information gets retyped, how long it takes to find out a status, what gets lost when someone goes on holiday. That page is half the analysis, and you can do it alone, before you talk to anyone.

The pricing page says where the platforms I build start and what moves the figure. And if you want a reference point for your own case, tell me what you need built: twenty minutes, after which I tell you whether it is worth doing, roughly how much and how long — or that I am not the right person for it.