What a website redesign actually costs — honestly

I would like to open this with an industry average. There isn't an honest one.

I went looking for the survey that everyone is implicitly citing when they quote a figure — a study that asked a representative sample of businesses what they actually paid. I could not find one. What exists instead is three things, none of which is that, and all of which get quoted as though they were.

That is an unsatisfying answer and it is the true one. The alternative is to repeat a number I cannot stand behind on the page where you are deciding how much money to spend, which would be a strange place to start being unreliable.

Three categories of figure circulate. Each is real; none is a survey of buyers.

Directory aggregates. Clutch, a B2B services directory, reports that "the average cost of web design projects is $38,105.39", drawn from reviews submitted by clients of agencies listed on its own platform.1 The same page also states that most web design projects reviewed there "usually cost less than $10,000". Both figures are from the publisher; the gap between them indicates an average pulled upward by a small number of large engagements, and the two-decimal precision is not meaningful on self-selected review data.

Practitioner surveys. The Web Designer Academy's 2025 pricing report is based on 208 survey responses from web designers, collected between 2 March and 4 April 2025, and records 57% of respondents pricing projects between $2,500 and $9,999.2 Its methodology is disclosed and its sample is small and self-selected. It measures what designers charge, which is a different quantity from what buyers pay.

Vendor price lists. The third category is agency pricing pages, sometimes aggregated and presented as research. These describe what a set of sellers advertise.

Figures that could not be traced to any of these are not reproduced here. One widely-circulated average, quoted to the pound with a year-on-year change attached, has no locatable publisher at all.

Since the averages are not much use, here is what actually changes a quote. In roughly the order they matter:

• How many pages have to exist, and how different they are. Twelve pages built from three layouts is a small job. Twelve pages that are all different is three times the work and looks identical on a proposal.
• Whether the content exists. Writing is the single most common thing that turns a four-week project into a twelve-week one. If the copy has to be produced, that is a separate piece of work and it should be priced as one.
• What the site has to talk to. A booking system, a CRM, a payment provider, an existing login. Every integration is a small project with its own failure modes.
• Whether the old addresses have to survive. Inventorying every existing URL and writing a redirect for each is real, billable work, and it is the part that protects everything the old site earned. Why that matters is its own article.
• How fast decisions come back. The one nobody prices. A project where feedback lands the same day finishes in a fraction of the time of one where it takes a fortnight, and the difference is entirely on the client's side of the table.

Anyone quoting before they know these five is guessing, including me.

Whatever the number turns out to be, some things are not upgrades. A site that works on a phone. Every old address redirected. The code and the domain in your name. A way to change a paragraph without raising a purchase order. A page that loads without asking a dozen other companies for permission first.

If a quote treats any of those as an optional extra, that is worth asking about before the price is.

The shape of the quote matters as much as the figure. A fixed number for a defined scope moves the risk of an underestimate onto the person who wrote it. A day rate moves it onto you, and it does so quietly — the number that gets agreed is the rate, while the number that gets paid is the rate multiplied by an estimate nobody has committed to. Neither model is dishonest. But only one of them tells you what the project costs before it starts.

A proposal worth reading itemises what is in scope, what is explicitly not, what happens when something is discovered mid-build, and what is included after launch. A proposal that is a single figure and a start date is not cheaper. It is just less specific about where the disagreement will happen.

Two things routinely turn up as extras and are worth naming early: who writes the words, and who owns the code and the domain at the end. The second of those has a default answer in law that surprises most buyers — it is its own article.

It would be odd to write all of the above and then be coy.

As a guide, a straight website build starts around €2,000, and a credibility rebuild — where the brand system is rebuilt alongside the site — starts around €3,500. A prototype sprint starts around €2,500. Custom software and full product builds are quoted case by case and start well above those figures, because the scope varies too much to shelf-price honestly.

Those are starting guides rather than a price list. Every project is evaluated on its own complexity and the work actually in it, and what you get before anything begins is one fixed number for your scope and your timeline. The three fixed-scope offers set out what each one includes.

If the five drivers above tell you this is a bigger job than that, they are probably right, and I would rather say so on a call than discover it in week three.