What custom web development costs in Greece

Nobody in this industry publishes prices, which is why every buyer in Athens has had the same experience: three quotes for one brief, arriving at four thousand, nineteen thousand and seventy thousand euros, with no way to tell which of the three understood the job. Here is the breakdown we would want if we were on the other side of the table.

The bands we work in are the ones on our own briefing form, and they mean something specific. Five to fifteen thousand is a small custom marketing site: a handful of unique templates, one language, a CMS the team can actually use, no integrations beyond a form and analytics. Fifteen to forty is the common case — a full site of ten to twenty templates, two languages, a proper content model, a couple of real integrations. Forty to a hundred is a platform: e-commerce with a custom checkout, ERP and warehouse in real time, or an application with accounts and roles. Above a hundred you are buying software with a website attached, and the conversation stops being about pages.

Inside any of those, the money divides in roughly the same proportions. Discovery, information architecture and the content model take about a tenth, and it is the tenth most often skipped by the cheapest quote. Design takes a quarter. The build takes something like forty per cent. Integrations take fifteen and are the least predictable line on the page. Quality assurance, accessibility, performance work and launch take the last tenth — the part that gets compressed when everything before it overran.

So why does one brief produce a fivefold spread? Four reasons, in order of how much damage they do. The first is that the cheap quote counted pages and the expensive one counted unique templates. Two hundred pages that share four layouts is a small job; twelve pages that all look different is a large one, and a quote priced per page is a quote from someone who has not read the design. The second is integrations: everyone lists them and few have asked whether the client's ERP has an API, who holds the credentials, and how long that company takes to answer. The third is who writes the copy and migrates the content — an enormous, invisible cost that gets assigned to the client in a footnote and then lands back on the project as a six-week delay. The fourth is the second language, which is cheap to plan for and expensive to add afterwards.

That last point generalises, and it is the single most useful thing to know before signing anything. Some changes are nearly free later: a new page, a new section, a colour, a campaign layout built from components that already exist. Others are structurally expensive: adding a language after content entry has started, adding a checkout to a site that was not built to hold accounts, changing the content model once editors have filled it, and adding a mobile app to a platform whose API was written to serve one website. Front-load the decisions that are expensive to reverse, and defer everything else. That is the whole discipline of not wasting money on software.

Then the year-two number, which almost never appears in a quote. Hosting for a Next.js site is between nothing and a few hundred euros a month depending on traffic and whether it sits on a managed platform or a droplet you own. A headless CMS is a subscription or a self-hosted instance somebody has to patch. Monitoring, error tracking and uptime alerting cost less than one bad morning. Then a maintenance retainer, which is where honest agencies and dishonest ones separate: it should cover dependency and security updates, performance regression checks and a defined number of hours for changes — not merely the right to send an email.

The counter-intuitive part is that this is where custom wins the argument it lost on day one. A theme's second year is spent fighting the theme: plugin conflicts after an update, a design change that cannot be made, a developer billing to work around a decision somebody made for ten thousand other businesses. A codebase written for one brand has none of that friction, so year two is mostly the changes you actually wanted. Cheaper to start, more expensive to live in, is the trade — and it is a good trade for a brochure and a bad one for a product.

One local note. Digital transformation grants under ΕΣΠΑ and the Recovery Fund have periodically covered a share of exactly this expenditure for Greek SMEs, and the terms change with every cycle. It is worth checking what is open before scoping, because eligibility often shapes the phasing of a project more sensibly than a preference would.

And a word on fixed price. We quote a fixed price for a scope that has been defined properly, because a client should be able to plan. What we will not do is fix a price to a brief that says integrate with the ERP without anyone having seen the ERP. That is not a quote, it is a bet, and the party that loses it is always the one who has already paid.