2026 Website Rebuild Costs in Australia: $100K or $1M

Three agencies quote the same brief. The numbers come back 2.5 times apart. Most people read that as one agency being expensive. It usually is not.

The gap opens up for a reason that sits further back than the tender. Before the brief, before the shortlist, someone had to put a number in a business case and defend it.

That person is usually the Head of Digital or the Head of Marketing, asked what a platform will cost before anyone has worked out what it needs to do. There is nowhere to look it up, because no agency in this market publishes its numbers. So they ask around, and every answer describes a different kind of project using the same word: website.

The number goes into the business case anyway, because the meeting is on Thursday. From then on, every decision is made inside a budget set before anyone knew what it was buying - and the quotes that come back later are that unmade decision, made several different ways by several different agencies.

The rule that should set the number

One question decides what a platform should cost, and it is not "what do these things go for?"

It is: what does this platform need to earn?

Almost nobody runs that calculation before going to market, and it is not complicated. How much revenue runs through this channel now. What a realistic lift in conversion, order value or enquiry volume is worth over 3 years. What the margin is on that.

Run it for a retailer doing $8 million a year through its site. In our platform work, a defensible lift from a proper rebuild - better journeys, faster pages, a checkout built to convert - is 10 to 15 per cent. Call it 12: roughly $2.9 million in extra revenue over 3 years, and at a 40 per cent margin, about $1.15 million the platform has to go out and earn. That number sets the budget, not what the last agency charged someone else.

To use that rule you need to know what the different numbers actually buy. So here they are.

What each number buys

These numbers come from our own platform work in Australia. They are not published research and we are not presenting them as a benchmark. They are what we have seen these projects cost, and what the money buys.

None of the 4 is the wrong answer. Each is right for a particular business. The money only gets wasted when a business buys one of them for a job that needed another.

Under $150,000

Right when the website is not a revenue channel

Plenty of good businesses sell through people, channel partners or a sales team. The website has a real job - look credible, load fast, be easy to find, be easy to keep up to date - but it does not carry the revenue.

At this level you get a proven platform, set up properly, with good design applied. For a business whose website is not carrying commercial weight, that is not a compromise. It is the right answer, and spending more is waste. No earning test applies, because the site is not being asked to earn.

What it will not do is run a revenue channel. There is very little connection to your other systems and very little room to grow into one.

$150,000 to $300,000

Right when the website drives direct revenue

People buy, book or apply on the site, and the business can put a number on the channel. The catalogue, booking flow or application process is well defined and does not need to reach far into the rest of the business.

The money goes into what moves that number. A design system rather than a set of pages. Real work on the journeys people take to buy, and the points where they give up. A checkout built to convert rather than merely function. Connections to payments, CRM and marketing tools. Measurement wired in from the start, so you can see which changes earn their keep.

What it will not do is reach into your operations. Stock, fulfilment, loyalty, point of sale, pricing rules that live in another system. Those connections are what the next level is for, and the reason it costs what it does.

$300,000 to $550,000

Right when deep commercial activity runs through the platform

The transaction is no longer self-contained. What a customer can see, buy or be offered depends on what is true in your other systems: stock, pricing, fulfilment, loyalty, membership. When the platform underperforms, someone notices in the numbers, and the cause is usually in that chain rather than on the page.

The platform is built around the systems you run rather than around a template, and connecting them - usually 3 to 8 - is the largest share of the work. Every connected system brings its own data model, its own way of failing and its own maintenance, so cost scales with how many systems talk to each other, not how many pages the site has.

It is also where customer experience stops being a design opinion and becomes an investment with a return. Journeys get tested with real users rather than assumed. Conversion gets engineered rather than hoped for. A single percentage point is worth real money to a business this size, and this is where the experience work is funded well enough to go and find it.

This is where most Australian mid-market rebuilds belong. It is also where most of them are underfunded, because the business case was written before anyone worked out what the platform had to earn.

$550,000 to $1 million and beyond

Right when the platform is the channel

The revenue runs through it. Ordering, payments, loyalty, identity, fulfilment - often across more than one brand, more than one region or a franchise network with its own politics and its own systems. The earning test is several million over 3 years, and businesses at this level usually already know their number.

The connections between systems are the project, at greater scale: 10 or more is normal, and the engineering is unforgiving because everything has to reconcile. Data and personalisation change what a customer sees. Experimentation runs continuously, because at this volume a small lift in conversion pays for a lot.

The experience work is the largest single driver of return at this level, not the smallest. When a platform carries the revenue, the difference between a good journey and an average one is measured in millions.

What this is not is a status purchase. If the website is a brochure with a contact form, spending here is not ambition. It is the same mistake as underspending, made in the other direction.

Why the quotes come back 2.5 times apart

When 3 agencies quote the same brief at $240,000, $410,000 and $620,000, they have not priced the same thing 3 ways. They have each priced a different thing, and the brief did not say enough to stop them.

One has quoted a build that stays close to the platform's standard behaviour. The shape of the work is known in advance, so the number is confident and low. Another has quoted a build that reaches into your systems, your data and the way your customers behave. More of that work is specific to you, so the number is larger. Both have answered the brief - they have answered different questions inside it. Compare the two as prices for the same product and you are not running a competitive tender. You are choosing between two different products without realising it, and calling the difference a saving.

The brief could have stopped this. Put the earning-power number in front of every agency on the list and there is no room for 3 products to come back as 3 quotes. The gap does not open because agencies read briefs loosely. It opens because the brief leaves a decision unmade, and each agency makes it on your behalf.

What AI has actually changed about these numbers

Most people going to market this year assume AI should have made all of this cheaper. The honest answer is that it has changed some parts a great deal and others hardly at all.

The biggest change is at the start. AI lets us build several working concepts quickly and put real options in front of real users in the first weeks, before the money is committed. Design decisions get tested rather than argued about, and the expensive kind of mistake - the one you only discover after launch - gets much less likely. Senior people become useful earlier too: their time goes into strategy and customer experience, where the return comes from, instead of into first drafts. Testing has moved as well. AI covers far more of the platform and finds the awkward cases before your customers do.

What AI has not changed is the expensive part. It cannot decide what the platform should do. It cannot design a system that stays reliable when traffic doubles, or when something it depends on goes down at four in the afternoon. Left alone, it produces something that works in a demonstration and falls over in production. The architecture, the engineering judgement and the accountability after launch are still human, and on a serious platform they are most of the bill.

A straightforward website really can be produced now for a fraction of what it cost in 2022, and anyone who says otherwise is protecting something. A platform carrying ordering, loyalty, payments and identity across a dozen systems costs close to what it always did, because the cost was never in the typing.

What it costs to spend too little

A business that spends $150,000 on a platform that needed $500,000 does not find out at launch. The site goes live. It looks right. Everyone is pleased.

The problem arrives a year or so later, when the business asks the platform to do something it was never built to do. Connect to the ordering system. Show different things to different customers. Support a second brand. Now there are 2 options, and neither is good: bolt extensions onto foundations that were never designed to carry them, and pay for it every year afterwards, or rebuild.

Most rebuild. We are partway through one now: a business that invested in a site that looked right and could not reach its ordering system. Eighteen months later it is paying for a second platform while the first holds the revenue back. It will have paid for 2 platforms and had the use of one, and the 18 months in the middle is the more expensive half.

That cost never appears on the comparison sheet. Procurement compares the bids against each other. Nobody compares them against what the business will need the platform to do in 3 years.

3 ways to bring down what this actually costs you

Not every saving is a downgrade. Two of these are against an agency's commercial interest, which is why you rarely hear them.

• Cut the number of systems before you cut the scope: connecting systems is where platform budgets go, and most organisations bring more of them to a project than the business case needs, because the systems exist and nobody wants to be the person who left one out. Every connection is a cost you keep paying, not a cost you pay once. Ask which will move a commercial number in the first year. Push the rest to a later phase, where they can be reviewed and, often, quietly dropped

• Find out whether you need a rebuild at all: a good deal of what gets blamed on a platform sits in conversion, search, content and how people arrive, and can be fixed on what you already own for a fraction of a rebuild. That is worth a proper diagnostic before the money is committed. Sometimes the honest answer is that the rebuild is not this year's project. That is the cheapest outcome available, and it is worth going looking for

• Build a design system, not a set of pages: honesty first - this one does not lower the quote. It costs a little more at the start and lowers what everything afterwards costs, which is the number that matters. A design system is a library of components the whole platform is assembled from - a button, a card, a product tile, each designed once, built once, tested once, reused everywhere. Without one, every new page is new design and new build, and your team is back in front of an agency every time the business wants something. With one, marketing puts a campaign page together in a day and it looks right because it cannot look wrong. A brand change is made once and appears everywhere. Accessibility and performance are solved in the component instead of argued page by page. It usually pays for itself inside the first year, then keeps paying

The question to take into the room

When the bids come back far apart, the instinct is to ask the expensive agency to justify itself. Reasonable, and wrong: the answer will be a list of features, and it will sound like a sales pitch whether it is one or not.

Ask every agency on the list this instead. Which of these platforms have you quoted us, and what happens to it in 3 years if we succeed?

A good answer is specific. It names the systems the quote does not connect, what connecting them later would cost, and where the platform runs out of road. A bad answer talks about scalability in the abstract and assures you everything is possible. The answers will not be similar, and that is the information the price spread was hiding.

If you are working out what your own rebuild should cost, tell us what the platform needs to do and what it needs to earn, and our team will tell you what we think it should cost and why. Talk to us about your project.