What does it cost to open a restaurant in Sydney?

Why one number is not an answer

A build budget opens with a header block, and every figure below it depends on what that block says. The block records project type, location, floor area in square metres, premium level, the estimated cost, and the average cost per square metre.

The last of those is the number operators ask for first. It is also the one that cannot be given first. Cost per square metre is computed at the bottom of the model, not entered at the top. It is the total divided by the area, arrived at after every trade has been priced against a specific site. Quoting it up front inverts the model: it takes the one figure derived from everything else and presents it as though it were an input.

Two of those five variables are properties of the concept and three are properties of the site. An operator can fix the first two before signing anything. The other three arrive with the lease.

A budget is a set of trade lines, not a lump sum

A builder who returns a single total has quoted a feeling. The model RagingMonk works against decomposes a build into individual trades, each carrying a unit cost and a share of the total, so the budget can be interrogated line by line rather than accepted or rejected whole.

Decomposition is what makes a quote comparable. Two builders pricing the same room will disagree, and the disagreement is only useful if it can be located: one is carrying more preliminaries, one has priced joinery to a higher specification, one has left mechanical out of scope entirely. Side by side at the line level those differences are visible. As two totals they are invisible, and the cheaper total wins for reasons nobody has examined.

The full breakdown, and what each line is actually carrying, is set out in how a venue fitout budget is structured.

What we are not quoting, and why

We have opened eight venues, so we know roughly what the rest costs, but it varies so wildly by site, format and condition that any figure we published would be misleading.

You will need separate budgets for the lease and bond, fitout construction, kitchen and bar equipment, furniture, POS and reservation systems, licensing and council approvals, initial stock, and enough working capital to trade through a slow first quarter. A fitout alone can run from a light cosmetic refresh to seven figures on a ground-up build.

Any agency that quotes you a total cost to open a restaurant without seeing your site and your lease is guessing. What we can do is tell you honestly whether the concept justifies the spend before you commit to it: that is what concept feasibility is for, and we have told founders to stop.

The approvals nobody budgets for

The build is the largest line and it is not the whole number. Two categories sit outside a builder's quote because they sit outside a builder's scope.

The consultants. A designer's scoping note on one project listed four appointments the operator had not allowed for: a town planner, to deal with change of use and approved seat count unless the landlord is carrying the development application; a BCA consultant; a hydraulic consultant; and a mechanical consultant. None of them appear on a shopfitter's quote, because none of them work for the shopfitter.

The certificates. Development consent, construction certificate and occupation certificate are three separate instruments, and they do not have to belong to the same party. On one project the landlord resolved the development application while the tenant carried the construction and occupation certificates. Which side owns which is a term of the deal, and it is cheapest to settle before signing rather than to discover during delivery. The sequence those certificates impose on an opening date is covered in what actually delays a venue opening.

Two ledgers, not one

A project that tracks only what it has spent cannot answer the question that matters, which is what it still owes. Two registers run alongside the budget, and they record different things.

Capital in. Every injection recorded against the person who made it, the amount, the date, the entity it went into and the method. On any venue with more than one contributor, this is the record that prevents an ownership argument two years later, when everyone remembers their own contribution accurately and nobody else's.

Payments out. Item, order number, date, payer, reference, order amount, amount paid, and balance outstanding. The last two columns are what earn the register. A budget line shows what a trade was priced at. Only the ledger shows what has actually left the account and what is still committed.

The question that can be answered for free

The cost cannot be quoted from a webpage. A different question can be, and it is the more useful one: whether the concept earns back what it will take to build.

That answer is produced in a fixed order. Audit, competitive landscape, search visibility and market sizing all complete before a single recommendation is written or a figure proposed. Discovery comes first, every time, because a recommendation made before the market is measured is a recommendation about a market nobody has looked at.

The output is not a price. It is a break-even expressed in the unit the venue will actually count: the number of covers, bookings or events required before the build has paid for itself. An operator who can picture that number can make the decision. An operator holding only a total cannot, which is why the missing figure was never the thing standing between them and a decision.

How the money is usually staged

Project engagements are milestone-based: typically 40% to start, 30% at midpoint, 30% on completion. Larger Foundry programmes can be structured monthly. Retainers invoice monthly. We do not ask for the full amount upfront.

For a launch, the brand and digital spend lands 6–9 months before opening, the campaign spend in the 12–20 weeks before the doors open, and the growth retainer begins at or shortly after opening.