What a Feasibility Model Can and Cannot Settle
What a Feasibility Model Can and Cannot Settle
General information for South Australian landowners only. This is not financial, valuation, tax or investment advice. What your project would cost, what it would sell for, what finance is available and what your land is worth are questions for a quantity surveyor, a registered valuer, your broker or lender and your registered tax agent against your actual proposal. Nothing here is a recommendation to proceed with or refrain from any development.
The number at the bottom is the least reliable thing on the page
Every feasibility ends with a figure. Owners read that figure, and treat everything above it as the working.
It is the wrong way round. The figure is an output of assumptions that were entered before the model did anything, and the model's contribution is not to know those assumptions — it is to show what happens to the answer when each of them is wrong.
This is not an argument against modelling. It is an argument about what the exercise is for. A model that produces a confident number and no visible assumptions has done the opposite of its job: it has hidden the very thing the owner needs to see.
This article is about the boundary of the tool. It is not about what should be in one — that is what a feasibility study should contain — and it is not about how margin is expressed, which is development margin on cost.
Where the fragility comes from
There is a structural reason a development appraisal is more sensitive than owners expect, and it is worth understanding rather than taking on trust.
Under the residual method, the value of the land is arrived at by assessing the value of the finished scheme and deducting all development costs, including profit and finance. Because the output is a residual, it can be highly sensitive to the inputs used, which is why sensitivity analysis is applied to show how minor changes to an input affect the result (source: RICS — APC: valuation approaches and methods).
That is the whole problem in one sentence. A residual is a difference between two large, uncertain quantities, so uncertainty in either one arrives at the answer magnified rather than diluted. A model can be arithmetically perfect and still produce a figure that moves a long way on inputs nobody thought were contentious.
Which is why a single number, presented alone, misrepresents the state of knowledge it came from. The honest form of the answer is a range with the conditions attached.
What a model genuinely settles
It is a real tool, and these are the things it does well. Owners under-use it in exactly these four ways.
It compares options under identical assumptions. This is its strongest use. Two schemes modelled on the same inputs can be ranked even where neither absolute number is reliable, because the shared errors largely cancel. Comparing forms, comparing selling now against developing, comparing staging — all of it is more trustworthy than any single result: subdivide or sell.
It finds the break point. Not "does this work" but "at what point does this stop working". That is a far more durable answer, and it survives being wrong about the starting figures.
It makes the assumptions explicit and arguable. Once a build rate, a sale rate and a programme are written down, they can be challenged by the people who actually know — a QS, a valuer, an agent. Undocumented optimism cannot be challenged, which is why it survives.
It shows where the sensitivity is concentrated. Some inputs move the answer a great deal and some barely at all. Knowing which is which is what tells you where to spend money on certainty — and that is the single most valuable thing a model produces.
What it cannot settle, no matter how good it is
These are not modelling failures. They are questions that belong to other people, and a model imports their answers rather than generating them.
What the finished product sells for, and how quickly. That belongs to a registered valuer and your selling agent. Absorption is usually the assumption with the longest reach: presales and absorption in Adelaide.
What it costs to build. A quantity surveyor (QS) or a builder, and it moves: build cost escalation and insolvency in a feasibility.
Whether approval will be granted, and on what conditions. No model has a view on this. Conditions can change the cost base after the model was signed off: the SA development approval process.
What is in the ground. Investigations answer this, and the answers arrive after commitment on most projects: uncontrolled fill and Class P sites and reactive clay and site classification.
What the bank will lend against. A valuation is a separate exercise with its own basis, and it does not have to agree with your model: low bank valuations on development.
When. Programme is an input, and delay compounds through finance and holding costs: the cost of delay.
Tax treatment. Structure and treatment belong to a registered tax agent and are not inferable from a construction model: GST and CGT on selling subdivided lots.
A model that appears to have settled any of these has not. It has recorded someone's guess in a cell and carried it to the bottom of the page with authority it never had.
The check most owners skip
There is a discipline worth borrowing from valuation practice: a residual result should be cross-checked against the comparable evidence of land sales rather than accepted on its own (source: RICS — APC: valuation approaches and methods).
The reason this matters to an owner is that it catches a specific failure. If your model says the land is worth considerably more than comparable land is selling for, the model is making a claim about the market, not reporting one. It may be right — a site can genuinely be worth more to a party who can do something others cannot. But that has to be stated as the argument it is and defended, rather than emerging quietly as arithmetic.
The same cross-check works when an offer is on the table: a model is one way to read whether a price is fair, and it is not the only one: is a developer's offer a fair price.
The failure mode: a model that stops being a tool
Almost every bad feasibility outcome has the same history, and it is not a spreadsheet error.
A model is built at the start, when the least is known. It supports a decision to proceed. Then the project starts producing real information — a QS estimate, a soil report, conditions of approval, a valuation, a tender — and the model is never re-run. It has quietly changed from a decision tool into a justification for a decision already taken.
The discipline that prevents it is unglamorous:
Re-run it every time a real number replaces an assumed one. Not to change the decision necessarily, but to know whether the decision would still be made.
Keep the assumption log visible, with a source and a date against each entry. An assumption whose owner cannot be named is not an assumption; it is a hope.
Watch the break point, not the bottom line. The useful question through delivery is how much room is left before it stops working.
Decide in advance what would change your mind. A threshold set before you are committed is worth far more than one negotiated with yourself afterwards.
Where the answer is genuinely uncertain, the right response is often to buy certainty on the input that moves the answer most, and re-run — rather than to proceed on the model or abandon on it.
Frequently asked questions
How accurate is a feasibility? Its accuracy is the accuracy of its inputs, amplified by the residual structure. A figure carried to the dollar is precise, which is a different thing from accurate.
Should I get a QS involved early? Cost is one of the inputs the answer is most sensitive to, and a QS is the person who can give it to you. What it costs on your project is for them.
My model says the project works. Is that enough to proceed? It tells you the project works if the assumptions hold. The useful follow-up is which assumption, if wrong, ends it — and how confident you are in that one.
How often should it be updated? Whenever a real number replaces an assumed one, and before any decision that is hard to reverse.
Can a model tell me whether to develop or sell? It can compare the two under the same assumptions, which is one of its stronger uses. It cannot tell you what the market will do, and it does not weigh your own circumstances.
Who decides what, and where Cyberate PM sits
What the finished product is worth — a registered valuer and your selling agent.
What it costs to build — a quantity surveyor (QS) or your builder.
Whether approval is likely, and on what conditions — your experienced planning consultant and the relevant authority.
What finance is available and on what terms — your broker or lender.
Tax treatment and structure — your registered tax agent.
Cyberate PM does none of those. We do not value property, price construction or provide financial advice. We work owner-side: we get real inputs from the right advisers into the model rather than leaving assumptions unowned, keep the assumption log current so the answer can be re-tested when reality arrives, and hold the decision against a break point set before commitment rather than after. What that role covers is set out in what a development manager actually does.
Before you rely on a feasibility
Which assumptions move the answer most — and who owns each of them?
Is the answer expressed as a range with conditions, or as a single figure?
Has the result been cross-checked against what comparable land actually sells for?
What would have to be true for this to stop working, and how far away is that?
When was it last re-run, and against which real numbers?
What have I decided in advance would change my mind?
A model does not tell you whether to proceed. It tells you what you are betting on — which is the more useful thing to know.
Market conditions, construction costs, finance terms and planning requirements change over time. Nothing here is advice about your project or a recommendation to proceed. Obtain your own valuation, cost, finance and tax advice before acting. Reflects publicly available material as at August 2026.
Sources
Expert property development and project management insights.
Not Sure Whether Your Site Stacks Up?
Send us the address and your goal. We will identify the first planning, buildability and feasibility questions before you commit further capital.