What a custom website actually costs in 2026

Two quotes for the same brief can land four hundred percent apart. That is not because one agency is cheating you. It is because the two quotes are usually for different things, and the brief was too vague to reveal the difference.

So here is what actually moves the price.

The ranges we quote

These are real numbers for our own work, converted roughly. Your local market may sit higher or lower.

Scope	Typical range	What it means
Focused business site, 5 to 7 pages	A$500 to A$2,500	A brochure site done properly. You supply hosting and domain at the low end.
Marketing site, 10 to 20 pages	A$2,500 to A$8,000	More pages, a CMS so you can edit content, structured data, analytics.
E-commerce store	A$3,000 to A$20,000	Depends almost entirely on how many systems it must talk to.
Custom web application	A$8,000 to A$50,000+	Logins, roles, dashboards, data that has to be right.
Custom software, ERP, CRM	A$15,000 upward	Typically months, not weeks.

Notice how wide those bands are. The band is wide because the scope is usually unclear at the quoting stage, and the honest answer is “it depends on these five things.”

What actually moves the number

How many decisions are already made. A client who arrives with a clear spec, approved content and a design they like pays far less than one who says “make it look good” and decides page by page. Discovery is not free. On a vague project, we might spend two weeks just establishing what we are building. That cost exists whether it is invoiced separately or buried in the total.

Whether content exists. Copywriting, photography and product data are real costs. If you have them ready, you save money. If you do not, someone has to produce them, and that someone bills by the hour.

How many systems must connect. A standalone site is simple. A site that must sync inventory with a warehouse, push orders into an ERP, and pull prices from a supplier feed is not. Every integration is a place where things break, so every integration needs error handling, retries and testing. This is the single largest cost multiplier on e-commerce work.

Whether users log in. Authentication changes a project’s category. Once there are accounts, you need password resets, permissions, session handling, and decisions about what each role can see. A site with 5,000 users and one with 50 need different architecture.

How much traffic you expect. A site expecting 200 visitors a day runs on modest hosting. One expecting 200,000 needs caching, a CDN, database tuning and monitoring. Building for scale you will not reach for three years is money burned, but so is rebuilding at month two.

How much you will change your mind. This is the honest one. Every agency has a story about the client who approved the homepage then asked for it rebuilt. Change is normal and expected within reason. Unbounded change is the fastest way to turn a fixed-price project into a painful one, which is why good contracts define what a revision is.

Why fixed price still matters

Hourly billing transfers all the risk to you. If the agency underestimates, you pay for their mistake. Fixed price transfers that risk to them, which is why they should be the ones scoping carefully before quoting.

We quote fixed after discovery for that reason. It forces us to understand the project before committing. If we get it wrong, that is our problem, not an invoice you did not plan for.

The one legitimate exception is genuine scope change. If you add a payment provider and two new languages halfway through, the price should move, and it should be quoted before the work starts rather than discovered on the final invoice.

What to ask any agency before you compare quotes

1. Is this fixed price, and what counts as a revision versus new scope?
2. Who owns the source code when the project ends?
3. What happens in month four if something breaks?
4. Which parts are custom-built and which are licensed third-party components?
5. What is explicitly not included?

That last question is where most budget surprises hide. “Not included” often covers hosting, domain, email, content, ongoing maintenance, and third-party licence fees. None of those are unreasonable to exclude. All of them are unreasonable to discover late.

The cheapest option is rarely the least expensive

A A$400 template site that cannot be extended, ranks poorly, and takes three seconds to load will be rebuilt within two years. Two rebuilds cost more than one careful build. If the site is genuinely a placeholder while you test an idea, a cheap build is rational. If it is meant to bring in customers for the next five years, the arithmetic changes.

We would rather tell you that now than sell you a rebuild later.