What a website costs in 2026, and why the range is so wide
Asking what a website costs is like asking what a house costs. The answer depends less on the number of pages than on how many decisions have to be made and who makes them. Below is the structure of the price, without the marketing.
What actually drives the price
Not the design, and not the page count. Four things decide it:
- Who writes the copy. If there is no content and the client expects the agency to produce it, that is separate work — often larger than the build.
- How many languages. A second language is not "twice the text". It is hreflang, separate URLs, a switcher that does not lose the current page, and an admin where both languages are edited side by side.
- Whether there is an admin panel. A static site is cheaper to launch and more expensive to live with: every copy change becomes a developer ticket.
- Whether it has to rank. The SEO frame — server-side rendering, structured data, canonicals, speed — is an architectural decision, not a line item bolted on at the end.
Three realistic tiers
Landing page, 1–2 weeks
One page, one language, client-supplied copy, an enquiry form. Good for testing demand. The risk: in six months it gets rebuilt, because nobody planned for a second language or a blog.
Corporate site, 3–6 weeks
5–15 pages, two languages, an admin for text and images, the SEO frame, a form with antispam and notifications. This is the most common request and the best ratio of price to useful life.
Site as a product, 2–4 months
A catalogue or blog with dozens of pages, search, user accounts, CRM or payment integrations. This is where queues, tested backups and monitoring appear — operations, not just development.
What vanishes from cheap quotes
When comparing two estimates, look for these lines in the cheaper one. Usually they are absent:
- A prototype before design (changes cost nothing on a prototype and a lot on built markup).
- Responsive work against real breakpoints, not "we will shrink it on mobile".
- Technical SEO: canonical, hreflang, sitemap, structured data, a performance budget.
- Accessibility: contrast, focus states, keyboard navigation.
- Form spam protection and an alert when the form breaks.
- Database backups with a restore test.
- Handover: credentials, documentation, the ability to hire a different team.
A cheap quote rarely means cheaper work. More often it means part of the work was not done, and nobody mentioned it.
Cost of ownership
Development is a one-off number. After it come hosting and the domain, dependency updates and security patches, monitoring, and content changes. That last one is where an admin panel pays for itself: if your marketer edits copy directly, you are not paying per comma.
A practical rule: budget 15–20% of the build cost per year for upkeep. Less than that and the site quietly ages until it breaks on an update that was deferred for two years.
What to ask before signing
- Who writes the copy, and is it in the price?
- Will there be a prototype before the design?
- What exactly can I change without a developer?
- What does handover look like — do I get the repository, the credentials and a manual?
- What is included in support, and what is the response time?
- What would a third language cost a year from now?
That last answer tells you whether the architecture has room. If a third language is "almost free", you are talking to someone serious. If it means "we would have to rebuild", you are seeing the real price — just deferred.
- Price is driven by content, languages, the admin panel and search requirements
- Prototypes, technical SEO and accessibility are the first things cut from cheap quotes
- Budget 15–20% of the build cost per year for upkeep
- Asking what a third language costs reveals how much room the architecture has
Usually a different scope: no prototype, no second language, no admin panel, no technical SEO. Compare the lists of what is included, not the totals.
Yes, if the first stage leaves room: server-side rendering, a sane URL structure and a place for a second language. Growing a quick landing page usually costs more than rewriting it.
