How Many Comparable Sales Support a Price per m²?

Behind every price per square metre lies a number of sales — but the figure that gets presented rarely says anything about how many. A report can show a neat weighted average of DKK 38,400/m² without revealing that it rests on three sales, one of which is an intra-family transfer and another an apartment twice the size of the property being valued. Three sales can look convincing on paper and still collapse at the first critical question from a credit committee.

The question of the right number of comparable sales behind a price per m² is not academic. It decides whether a valuation can withstand a softening market, a complaint, or an auditor who wants to see the basis. And the answer is rarely a fixed number — it depends on how homogeneous the segment is, how widely you have searched, and what the individual sales actually tell you. Here is how to judge whether the basis holds.

Why the number alone is not the answer

It is tempting to treat the count as a quality marker: the more sales, the better the valuation. But the relationship is not linear, and it breaks down at both ends.

At the low end the problem is obvious. One sale is anecdote — it tells you what one buyer paid for one property on one day, and it can be driven by anything from a divorce sale to a buyer with a personal attachment to the address. Two or three sales do not give you an average; they give you a random sample so small that a single outlier can move your price per m² by several thousand kroner.

At the high end the trap is more subtle. If you have to gather 100 sales to arrive at a “robust” figure, you have probably widened the search so much in radius, time or property type that you have mixed apples and pears. A large count can dilute rather than strengthen: a precise valuation on 15 genuinely comparable sales is worth more than 100 sales where half sit in a different price segment. The count only becomes meaningful when it is read together with the dispersion — two figures that always belong together.

Rule of thumb: never report a price per m² without also stating the number of sales behind it and the dispersion between them. A figure without those two context measures is not a valuation, it is a claim.

What dispersion tells you that the count cannot

Picture two valuations, both based on ten sales. In the first, the ten prices per m² fall between DKK 36,000 and 40,000 — a dispersion of around 10%. In the second, the same ten sales spread from DKK 28,000 to 52,000. Both produce the same average, but the two bases are not in the same league.

The low dispersion tells you the segment is homogeneous and that the ten sales genuinely describe the same market. The high dispersion tells you something is wrong: either the properties are not comparable, or the selection spans several sub-markets that ought to be separated. A high number of sales cannot rescue a high dispersion — it can only mask it inside an average.

That is why it is more robust to look at the median than the average when the dispersion is large: the median is not affected by a single extreme sale at each end. And it is why you should always be able to answer how many of the sales fall within ±10% of your conclusion. If the answer is “two out of ten,” the count has fooled you.

A working basis for how many it takes

There is no statutory minimum, and any fixed number should be taken as a starting point, not a rule. But as a practical bearing for an ordinary residential segment:

• Fewer than 5 sales: Insufficient as a standalone basis. Can be used indicatively, but the conclusion must be flagged with reservations and supplemented by professional judgement.
• 5–10 sales with low dispersion: A usable basis for a homogeneous segment where the properties genuinely resemble one another.
• 10–20 sales with low dispersion: A solid basis that can support a valuation with high confidence.
• Many sales with high dispersion: A warning sign, not a quality stamp — the segment should probably be narrowed.

The numbers shift with the segment. For atypical properties — large detached houses, decommissioned farms, commercial conversions — you will rarely find 10 genuinely comparable sales, and here professional judgement necessarily carries more weight. The point is not to chase a magic number, but to assemble a basis made up of sales that are genuinely comparable. What that involves we have covered in our review of what makes two properties genuinely comparable, and it is the precondition for the count meaning anything at all.

The number of sales and the confidence level go together

A valuation is rarely just a number — it should come with a confidence level that tells the recipient how secure the basis is. And here the number and dispersion of comparable sales is the primary driver.

A well-defined confidence level can be built on three inputs viewed together:

1. The number of genuinely comparable sales in the selection.
2. The dispersion between the individual prices per m².
3. The currency — how many of the sales fall within a relevant time horizon.

When all three are strong — many sales, low dispersion, fresh data — you can report high confidence. When one of them is weak, the level drops, and that must be reflected explicitly. A valuation on three sales with high dispersion is not “wrong,” but it is uncertain, and that uncertainty belongs to the recipient — the bank, the credit committee, the client — not just to your own gut feeling. Hiding it is the opposite of professionalism.

The currency and breadth of the search are a discipline of their own, because there is always a trade-off: the wider you search in radius and time, the more sales you get — but the less comparable they become. We have treated that balance in our article on radius and time horizon for comparable sales, and it is inextricably linked to the question of number: you can always obtain more sales by relaxing the criteria, but the price is quality.

How to document that the count holds

A sufficient number of sales is only half the job — the other half is showing it. A valuation that concludes on a strong basis but only reports the final price cannot be told apart from one built on three random sales. The difference lies in the documentation.

In practice that means the report must be able to answer:

• How many sales go into the basis?
• What is the dispersion, and how many fall within ±10% of the conclusion?
• Which sales were deliberately excluded — and why?

It is precisely the exclusions that matter. If you have looked at 30 sales and used 12, it is the 18 you set aside — intra-family transfers, divergent areas, forced sales — that show the count has been quality-assured rather than being a raw total. How to sort them out we have covered in our guide to selecting comparable sales, and how to describe the whole chain transparently for a critical reader in our guidance on documenting the comparables in a valuation report.

Remember at the same time that transaction data from the market must be read together with the public registers. An area figure from BBR (the Buildings and Dwellings Register), a local plan from Plandata.dk or an easement in tingbogen (the Land Registry) can explain why a sale in the selection diverges — and so help you decide whether it belongs or should be sorted out. The count only holds when each individual sale is understood.

When the manual exercise becomes automatic

Gathering a sufficient number of comparable sales, calculating the dispersion, sorting out the outliers and translating it into an honest confidence level is a careful manual discipline — the one that separates a durable valuation from a quick average. That is exactly the exercise that Arcili’s Boligvurdering (Valuation) automates: the model weighs up to 72 factors against comparable sales in a defined area and assigns one of three confidence levels, directly reflecting the number, dispersion and currency of the underlying basis.

It does not replace professional judgement — it gives you the basis transparently and documented, so you can spend your time assessing the divergences that genuinely call for a professional view. If you want to see how the confidence level is calculated on a specific property, you can book a walkthrough.