What's Another Square Metre Worth? Marginal Value vs Average Price

What's Another Square Metre Worth? Marginal Value vs Average Price

General information for South Australian landowners and small developers only. This is not valuation, planning, legal or financial advice, and nothing here is an opinion of the value of any particular property. If you need a figure you can rely on for a transaction, lending or a statutory purpose, that is a valuation, and the place to obtain one is a certified practising valuer. Route questions about what your site can physically and lawfully accommodate to a planning consultant; questions about sale contracts, disclosure and settlement to an SA property lawyer or conveyancer; and questions about how a development or sale would be taxed in your hands to a registered tax agent. Market conditions and statutory arrangements change over time, so confirm the current position for your own property before relying on anything here.

The number the internet hands you for free

Look up any Adelaide suburb and a median price appears before you have finished typing. It is free, it is current, it is quoted to the dollar, and it is treated as the suburb's value the way a thermometer reading is treated as the temperature. For a homeowner idly wondering how the street is travelling, it is harmless. For anyone deciding what to build, it answers far less than it appears to, and the reason is worth stating precisely.

A median is the middle transaction in a blended pile. Into that pile go big allotments and small ones, renovated houses and original-condition ones, new builds and old stock, homes on main roads and homes on quiet ones. The median tells you where the middle of that mixture landed. It does not tell you why. Two suburbs can share a median for entirely different reasons — one because its houses are modest on generous land, the other because its houses are generous on modest land — and a median can move without any underlying value changing at all, simply because the composition of what happened to sell shifted between periods. The statistic answers the question "what did the middle sale settle at", and that is rarely the question a development decision turns on.

It is worth marking this article's edges before going further. Whether a developer's offer for your land is fair is not answered by any per-square-metre figure — that question belongs to a different tool entirely, worked through in how residual land value frames a developer's offer. Whether you should subdivide at all, rather than sell as-is, is a decision with its own structure, taken up in the subdivide-or-sell question for Adelaide landowners. This piece sits underneath both: it is about what the market evidence can and cannot tell you about the worth of land and building area at the margin, because that is the evidence every one of those decisions eventually leans on.

The question a product decision actually turns on

Here is the question a developer is really asking, whether or not it is ever said aloud: if I put another square metre of land under each dwelling, or another square metre of floor area inside it, does the sale price move by more than that square metre costs me to provide?

That is a marginal question, and it is a different species from the average one. The average tells you what the whole bundle — land, building, condition, position — fetched in the middle case. The marginal question asks what the market pays for the next increment of one ingredient, holding the others still. And the two can point in opposite directions. A suburb can carry a healthy median while paying almost nothing at the margin for extra land, because buyers there are purchasing the dwelling and treating the yard as incidental. Another can pay firmly for land and shrug at extra floor area, because the buyer profile values the allotment more than the building that sits on it.

Every product decision on a development site is secretly a position on this relationship. Fewer, larger dwellings on more land each, or more, smaller dwellings on less — that is a bet on the land-to-building ratio the local market rewards. Whether the upper floor earns its construction cost, whether the third bedroom does, whether the courtyard should give way to the living room: each is a claim that the marginal square metre in question is worth more in one place than another. The median cannot arbitrate any of these claims, because it blends together sales that took opposite positions on all of them.

The marginal relationship can be estimated, but only from evidence that separates the ingredients — sales read against their land areas, building areas, ages and conditions, so that the contribution of each can be distinguished rather than averaged away. That is modelling work, and its output is not a price. It is a relationship: how settled prices in this market have responded to increments of land and of building, across the range the evidence covers.

The relationship holds over a range, and the range matters

That last clause carries more weight than it looks like carrying. A modelled relationship is only as good as the evidence underneath each part of it, and the evidence is not spread evenly. The middle of a market is thick with transactions — allotments and dwellings of the sizes people usually build and usually buy. Out in the tails, sales thin out. Very small allotments, very large ones, unusually compact dwellings, unusually sprawling ones: each trades rarely, and the few sales that exist are more likely to be odd in other ways too.

The consequence is that the marginal relationship is well-supported where the evidence is dense and increasingly speculative where it is sparse. What the market pays for another square metre in the heart of the observed range is a finding. What it would pay far beyond the largest configuration anyone has recently built is an extrapolation — a guess wearing the model's clothes. An honest analysis says out loud where its evidence runs thick, where it runs thin, and at what point the line on the chart stops being a summary of sales and starts being a conjecture about them.

This is not a defect to be apologised for. It is information. If the configurations you are weighing sit inside the well-evidenced range, the analysis can carry real weight in your decision. If one of them sits in a quiet tail, the correct conclusion is not that the option is bad — it is that the market has not spoken clearly about it, and the decision leans more on judgement and less on statistics than you might have hoped. Knowing which situation you are in is itself worth paying for.

Evidence hygiene: what the sales have to be before they count

None of the above is better than the sales it is built from, so it is fair to ask what discipline the inputs deserve.

Settled sales, first — prices at which contracts actually completed, not advertised prices, not agent estimates, not the figure a listing was withdrawn at. Advertising expresses hope; settlement expresses agreement.

Cleaned, second. Raw sales records carry passengers that do not belong in market evidence: transfers between related parties, sales of part-interests, transactions bundled with something other than the land and dwelling. That last category deserves particular care, because a sale that included an existing approval may not be evidence of the land alone, and leaving it in can distort the apparent margin. How much of such a price is attributable to the approval is a valuation question for a certified practising valuer, not something to settle inside the dataset. Whether an approval itself adds value — and when it can subtract it — is its own question, examined in what a DA-approved block is really worth; for present purposes the point is narrower: such sales must be identified and either adjusted or excluded, and the analysis should say which it did. Outliers handled silently are outliers hidden; outliers handled openly are part of the method.

Third, and least comfortable: no forecasting. A relationship estimated from settled sales describes what the market has done, over the period the sales cover. It does not describe what the market will do by the time your project reaches it. Treating a historical margin as a promise about a future settlement is the same error as treating a median as a value — a statistic pressed into service on a question it never answered. The honest use of the relationship is comparative and present-tense: given what the evidence shows, which configuration has the market rewarded, and how confidently can we say so.

When the market's favourite configuration does not fit your block

Suppose the analysis is done well and it points somewhere clear: in this market, the evidence rewards a particular balance of land and building. Then you lay that configuration over your actual allotment — its dimensions, its frontage, its easements, its overlays — and it does not fit. The statistically attractive product may be more than your site's dimensions and the planning criteria applying to it can support — something your planning consultant would need to confirm.

It is tempting to read that moment as failure: the data named a target and your block missed it. That reading is wrong, and the wrongness matters. A marginal relationship estimated from other people's sites carries an unstated condition — it applies to configurations that can actually be delivered. Where yours cannot deliver one, the statistic is not wrong; it is inapplicable. What you have learned is real and useful: the constraint is speaking, telling you which part of the market's preference structure your land can reach and which part it cannot. The feasible set comes first, and the evidence is then read over the feasible set — not the other way around.

This is also where market analysis hands over to other disciplines, and should say so plainly. Whether a given configuration can be arranged on your allotment, and what planning pathway it would attract, is a question for your planning consultant, with the relevant authority determining the application in fact — any concept scheme is subject to planning consent, however well the numbers behind it read. In research we co-authored on South Australian planning applications, a pattern worth carrying into this discussion was how much of what decides a site's realistic options is visible before anything is lodged; the market evidence and the constraint evidence are both knowable early, and the useful work is reading them together. And once a feasible configuration is chosen, execution still separates outcomes — dwellings that look alike on a spreadsheet can be received very differently by buyers, for reasons taken up in why similar homes sell differently.

Reading the margin on your own site

If you want this done systematically rather than by suburb median and instinct, this is what our market analysis report exists for. It builds market value indicators from cleaned settled sales in your area — indicators, deliberately, not a valuation, which is work for an appropriately qualified valuer — and returns an evidence-based assessment from historical cases of how the market there has rewarded land and building area at the margin, with the well-evidenced range and the thin tails disclosed rather than smoothed over. Where it compares configurations for a site, it compares them side by side with the evidence behind each; it does not auto-rank them, any concept scheme in it is subject to planning consent, and any recommendation it carries is signed by a named development professional rather than generated by a model — and that recommendation is limited to project-delivery matters, with the financial and valuation questions left to your own advisers. It is not a forecast of any future price, and it does not replace the valuer, planner or tax professionals who advise you in their own lanes. You can read what it covers at the market analysis report. Where the market read has to be weighed against build cost and the planning position together, that is the brief of property development feasibility. And if you would like a view on whether your situation warrants one, start a conversation with us — a conversation is a practical way to test whether the settled-sales evidence near you is deep enough to support the analysis.