What actually drives the cost of a website

Ask five agencies to quote the same website and the numbers will not be close. That is confusing enough that many businesses default to picking the cheapest, which is usually how you end up paying twice.

The spread is rarely about greed. It is that “a website” describes anything from a five-page template fill-in to a bespoke system. Here is what actually moves the number.

The things that genuinely drive cost

1. How many kinds of page there are

Not the page count — the number of distinct templates. Twenty blog posts using one template is far less work than five pages that each need their own layout.

When you brief a project, count templates, not pages. It is the single most useful number you can give someone quoting.

2. Whether the content exists

This is the most underestimated cost in the entire project, and it is usually yours rather than the agency’s.

If copy, images and product information already exist and are approved, the build moves fast. If they need writing, sourcing, editing and sign-off, the timeline stretches and someone has to do that work. Projects stall on content far more often than on code.

3. Custom design versus a template

A theme costs less because the design decisions are already made — and it shows, because your competitor may have bought the same one. Custom design costs more because someone is solving your specific problem: your services, your audience, your conversion path.

Neither is wrong. A template is a reasonable choice for a business that needs to exist online cheaply. It is a poor choice for a business whose credibility depends on looking better than its competitors.

4. Functionality beyond pages

Anything that stores, calculates or connects adds real cost:

• Booking or appointment systems
• E-commerce, payments, tax and shipping rules
• User accounts and logins
• Multi-language content
• Integrations with a CRM or existing software

Each of these is a system with its own edge cases, not a page.

5. Whether performance and accessibility are actually done

This is where cheap quotes quietly save money, because you cannot see it in a screenshot.

Doing it properly means optimised images, controlled font loading, restrained JavaScript, semantic markup, keyboard support, tested colour contrast and real device testing. Skipping it produces something that looks identical on the demo and performs worse in search, converts worse on mobile, and excludes some of your users.

6. How many rounds of revisions are assumed

An open-ended revision commitment gets priced accordingly, whether or not you use it. Structured rounds cost less because they are predictable. Agreeing direction before design starts is what keeps revisions few, and it is worth the extra conversation up front.

What does not deserve to add cost

• Page count alone, once the templates exist
• Stock imagery, which is usually a small licence fee
• “Mobile responsive”, which is not a feature — it is the baseline
• Basic SEO setup, which should be part of building the site properly, not an upsell

If a quote itemises “make it work on mobile” as an extra, that tells you something about the standard being applied.

How to compare quotes that look wildly different

Ask every quote the same five questions:

1. How many distinct page templates are included?
2. Who is writing the content, and is that in the price?
3. Is the design custom or based on a theme?
4. What happens after launch — who fixes what, and for how long?
5. Who owns the domain, hosting and the files at the end?

That last one matters more than people expect. If the answer is anything other than “you do”, be careful. Some cheap builds are cheap because you are effectively renting a site you can never move.

A reasonable way to think about budget

A website is not a one-off purchase, it is infrastructure. The question worth asking is not “what is the cheapest way to get a website” but “what does this site need to do, and what is that worth to the business?”

If a single new client is worth a meaningful amount to you, a site that generates a handful more per year pays for itself quickly. If it generates none, then whatever you paid was too much.

That is the real test, and it is why we would rather scope honestly than win a project on a number we cannot deliver well.