Two quotes are open on your desk. They describe roughly the same website, the numbers sit close enough together that price alone settles nothing, and the thing you actually need to know is which of them will still be true in month three.
A quote is a price for a scope, and a scope is a set of decisions somebody has already made. Where a decision is still open, the number prices an assumption about how that decision will land. That mechanism sits behind almost every website invoice that ends up above the estimate, and it explains the other direction just as well: the jobs that come in on the number are the ones where somebody could answer a question in the same week it was asked.
This piece is about what moves a price once the work is running. If what you want is the price itself, what a company website costs in Austria has the figures, the running costs and the items Austrian law adds on top.
Why does a website cost more than the quote?
A website costs more than the quote when part of the scope was still undecided at the moment somebody put a number on it. An estimate prices work that has been described. Anything not yet described gets priced as an assumption, and when the assumption turns out to be wrong, the difference arrives later as extra work.
Written down that looks obvious. On the quote itself it stays invisible, because an open decision sits on the page looking like any other line item. “Contact form” describes something already settled. “Booking system” depends on whether anyone has said which system the bookings land in, who reads them, and what happens to one made at ten on a Sunday night. Both take up a single line.
So a quote does not read usefully from top to bottom down the price column. It reads line by line, with one question against each entry: does this describe something we have already decided, or something we still have to work out? The second group is where the number can move.
What is a quote actually built from?
An estimate has inputs, and they are less mysterious than they look. Before we can put any number at all on a website, we need three things: how big the site is, how much traffic it carries, and how complicated it is underneath. After a thirty-minute conversation covering those three, a rough number is usually possible.
- Size. How many pages, and how many of them are genuinely different from one another.
- Traffic. What the site carries today, because that decides what the hosting and the caching have to survive.
- Complexity. What sits behind the pages: forms, second languages, a shop, a booking flow, anything that has to talk to another system.
A conversation is enough for that. A provider who can give you a range after half an hour has understood the shape of the job. One who can say nothing at all before a three-week discovery phase is either working on something genuinely complex or is still unsure what the job is.
What happens to an input that has no answer yet is the whole subject of this piece. It does not stop the estimate. It gets an assumption instead.
Which part of this is ours to carry?
Part of the gap between an estimate and a final invoice belongs to the provider, and it has no business being passed down.
We have priced projects and then found we had underestimated them. Mostly that was stakeholder management. Sometimes it was an integration, where a third-party system turned out to be far more complicated than it looked from outside. That is part of the game, and we stand by our word.
Estimating is one of the things you are paying a professional to be good at. The risk of being wrong about your own trade sits with the person who practises it. A provider who quotes low to win the work and then rebills the difference as unforeseen complexity has moved their own estimating risk onto a business that had no way of pricing it, which is worth checking for before you sign anything. How to tell whether a web provider is good goes through the rest of that check.
One honest exception belongs here. Work you order on top of the agreed scope is new work. That is a different thing from an estimate being wrong, and Austrian law treats it as a different thing too, which the legal section below comes back to.
What does a website project end up waiting on?
Four things, and none of them are technical. Each one is a decision, and each one has an owner.
| What stays open | What it looks like while the job runs | What settles it |
|---|---|---|
| Who signs off | Every round goes to a wider circle than the one before | One named person decides, everyone else advises |
| What comes across from the old site | Nothing gets dropped, so everything gets rebuilt | Decide what does not move before the build starts |
| Which systems have to connect | The booking tool works differently than its documentation suggests | Name every third-party system in the first conversation |
| What the site is for | The design gets discussed and nothing ends the discussion | One sentence about what a good week on this website looks like |
The third row is the one where the risk stays with us. If we said we could connect your booking system and its interface turns out to be from another decade, our reading of it was wrong and we carry that. The other three are decisions only your business can take, which is why they are worth taking before the build.
Now the part that gets underrated. Read that table again as a description of a large organisation: a sign-off circle that widens with every round, an archive nobody has permission to delete from, systems owned by four different departments, and a purpose that seven people would each define slightly differently. A business of five to fifty people usually has none of that. There is a person who decides, and they are in the room. That short decision chain is a real commercial advantage on a website project, and it is among the first things a growing business loses.
What does Austrian law say when an estimate is exceeded?
Austrian law answers this directly, and the answer does not depend on how the conversation goes. § 1170a of the Austrian Civil Code (ABGB) governs cost estimates and splits them into two kinds.
Under Abs 1, where the contract rests on a cost estimate given with an express warranty of its correctness, the provider may not demand any increase in the fee, even if the estimated work turns out to be unexpectedly large or unexpectedly expensive. Under Abs 2, where the estimate carries no such warranty and a considerable overrun proves unavoidable, the client may withdraw from the contract against reasonable payment for the work already done. In that second case the provider has to notify the client without delay as soon as such an overrun proves unavoidable, and loses every claim for the additional work if that notification does not happen.
| The estimate you were given | What may be charged above it | What has to happen if the work runs over |
|---|---|---|
| With an express warranty of correctness | Nothing, even if the work turns out larger or costlier than anyone expected (§ 1170a Abs 1 ABGB) | Does not arise |
| Without a warranty | The overrun, if it is objectively justified and unavoidable, and only if it was notified | The provider must tell you without delay as soon as a considerable overrun is unavoidable, or lose every claim for the extra work. You can then agree to it, or withdraw and pay reasonably for the work done (§ 1170a Abs 2 ABGB) |
Three practical things follow from that.
The first is the question to ask when a quote arrives: is this estimate given with a warranty of its correctness, or without one? With the warranty, the number is a ceiling. Without it, the number is a professional’s best figure carrying a notification duty. Both are legitimate, and both arrive in the same kind of document, which is why the difference only becomes visible once the work runs over. Ask anyway, because between businesses the WKO guidance says a cost estimate counts as non-binding in case of doubt, unless something else was agreed. A quote that says nothing about the warranty is therefore already the second case, not an open question.
The second is the notification duty itself, because it is stricter than most people expect. It bites at the moment a considerable overrun becomes unavoidable, not at the moment the invoice is written. A provider who stays quiet and presents the extra at the end loses the claim to it. On what counts as considerable, the WKO guidance on cost estimates between businesses (last updated 15 October 2025) reports that case law requires overruns of more than 15% to be notified to the client without delay. The same page states immediately afterwards that the 15% mark is only a guideline, and that what counts in the end always depends on the circumstances of the individual case. The figure comes from the courts rather than from the text of the statute, and it is a guideline rather than a threshold.
The third is a limit worth knowing before you rely on the first case. The same WKO page states that where the client has caused extra costs through change requests or additional orders, those costs may be charged on top in any event, including under a binding cost estimate. The warranty covers the provider’s estimate of the work you agreed. It does not extend to work you added afterwards. That is the mechanism from the sections above showing up in the law: a decision taken late reaches the project as a change request, and a change request is chargeable whatever the estimate said.
Two boundaries on all of this. § 1170a governs the overrun of an estimate a provider gave you, and says nothing at all about the level of the price itself. And a cost estimate is neither a fixed price nor a lump-sum price, because the ceiling in Abs 1 applies only where the warranty of correctness was expressly given. These are also the business-to-business rules, which are the ones that apply to your company.
What should you settle before the work starts?
Five decisions. None of them needs more than an afternoon, and all five are cheaper to answer now than in week six.
- Who decides. One name. The team advises, the named person signs off. Say it out loud at kickoff so the provider knows whose approval ends a round.
- What does not come across from the old site. Walk the current site and mark what ends with it: old service pages, three generations of team photos, the news archive from 2019. Anything you do not mark gets rebuilt, and rebuilding is chargeable work.
- Which systems have to connect, by name. The booking tool, the newsletter, the shop, the export into accounting. Name them in the first conversation, with the version you actually run, so the interface can be looked at before the number is written rather than after.
- What a good week on the new site looks like, in one sentence. Name the thing you want more of and roughly how much of it would count as working. That sentence ends design arguments that otherwise run forever, and three questions nobody asks before ordering a website covers the two others that belong in the same conversation.
- Whether the estimate carries a warranty of its correctness. Ask directly, and get the answer in writing, because the whole section above turns on it.
If you also want the sequence of a build, from the first conversation through to launch, getting a business website built in Austria walks through it stage by stage.
When does the question not come up at all?
A monthly fee takes the question away, because there is no project estimate for anything to exceed. Design, hosting, changes and support sit inside one fee that stays where it is when a decision arrives in week six instead of week one. Our plans start at € 295 per month, and the page lists what each tier covers.
That shape does not fit everything. A complex application, a large shop, or anything that needs engineering of its own still belongs in a project with an estimate. If that is what you are buying, the two questions from this piece are the ones that do the work: which lines on this quote describe a decision we have already taken, and does this estimate carry a warranty of its correctness?