An AI sales valuation of your home is based on fixed, documentable information about your home, not on guesswork or gut feeling. The information comes from public registers and objective features of the property. The point of an AI sales valuation is that its basis can be looked up and checked.
There are many sources. An AI model typically draws on BBR (the Danish building register), CVR (the Danish business register), tinglysningen (the land register), the energy label register, Vurderingsstyrelsen (the Danish Property Assessment Agency), Plandata, DAR (the Danish address register), the cadastral register and property tax information. Each source supplies its part of the picture, and together they cover the building, the ownership, the location and the public assessments of the property.
BBR is the key one. The register holds the property’s floor area, building type, age, mix of materials and location. These are the same fields used to set the property value and calculate property value tax. The floor area carries particular weight, because it feeds directly into both the valuation and the calculation of what the home costs to own.
The sale prices come from the land register, and they are filtered. Sales that do not reflect the market price are weeded out, such as transfers within families and forced sales. What remains are the sales that genuinely say something about what homes in the area have sold for.
Once the data is in place, an AI valuation can estimate a property’s value in seconds. It uses features such as floor area, floor level, year of construction, distance to public transport and price trends in the neighbourhood.
The public registers behind the answers: BBR, the land register and Vurderingsportalen
The answers an AI home valuation gives are based on public registers. Several of them you can look up yourself and correct if something is wrong.
-
BBR is updated continuously as building applications are processed, and keeping the data up to date is the job of the municipalities. If you have spotted an error yourself, you can correct your BBR details via ‘Ret BBR’ (correct BBR) by logging in with NemID or MitID.
-
On Tinglysning.dk you can view the unofficial land register certificate (tingbogsattest) for your own property free of charge using MitID. The certificate shows the deeds and easements that form part of the basis for the valuation.
-
Vurderingsportalen is run by Vurderingsstyrelsen under Skatteforvaltningen (the Danish Tax Administration) and is the access point for the public property assessments.
-
OIS (Offentligt Informations System, the public information system) gathers official property master data from all the relevant authorities, and public property and land values can be shown on interactive maps.
How the model calculates: GAM, KNN and uncertainty ranges
The public property assessment system does not use a single calculation. The model combines two methods: Generalised Additive Models (GAM) and K-nearest neighbour methods (KNN). Each has its own job, and the interplay between them gives an answer to what a home is worth.
The GAM model learns from historical data. It estimates how a property’s features affect the price, and it does so using flexible regression splines. That is because the price per square metre does not rise in a straight line with floor area. An extra square metre does not necessarily make the same difference to the price whether the home is small or large, and GAM can capture curves like these instead of forcing the data into a straight line.
The KNN component captures the local picture. It compares each property with the most similar reference properties, typically those nearby, and sets a local price level based on them. The calculation starts from a base price built on the average price in a given area, and adjusts it for differences in characteristics such as size, location and type.
Vurderingsstyrelsen adds further layers on top. Geographical data such as coordinates, zoning status and planning conditions is included, as are market figures such as rent levels and required yields for commercial properties, along with historical public assessments. For commercial properties, data on rental income and operating costs is also used in a yield-based model.
When the assessment is ready, the owner receives an assessment notice. It contains the key information behind the figure, so you can see what it was calculated from:
-
Which ‘standard home’ and area price were used as the starting point
-
Which data about the property was used
-
The uncertainty range for the assessment
An AI valuation of a home’s price is an estimate, and the uncertainty range shows how wide an interval the valuation falls within. It helps keep expectations in check, whether you use the public assessment or another price indication as your benchmark.
What the algorithm cannot see: condition, view and the association’s finances
An AI valuation has its limits. The model works with what can be measured and looked up, but a number of factors that matter a great deal to what a buyer will actually pay are beyond its reach.
-
The home’s condition, the view, noise from neighbours or the quality of a renovation are invisible to the model. It does not know whether the kitchen is from IKEA or Bulthaup.
-
For cooperative housing (andelsbolig), the model cannot assess the financial health of the housing association. This covers things like the maintenance plan, liquidity or the mix of loans.
That does not mean an AI valuation is useless. It is a quick indicator of general price trends and the market level in an area, and it is available straight away without booking a visit, but it does not replace a professional estate agent valuation, where someone has seen the home and knows the circumstances around it.
There is also a formal layer worth knowing about if you own a cooperative home. The maximum price of the individual home is calculated as the share value (andelskrone) × the home’s share + the value of individual improvements − depreciation. The maximum price is a ceiling: you may sell below it, but never above it. If you exceed the ceiling, it can lead to a claim for repayment.
The homeowner’s own answers: energy label, condition and documents
When the public data does not capture everything, you can fill in the gaps yourself. Many valuation tools let you adjust the estimate with information about the home’s condition, energy label and the value of its features.
The authorities also collect correct information from owners to update BBR, for example on floor areas and building details, so that the data gives an accurate picture. If you have corrected an error in your BBR details yourself, that is exactly the kind of input that makes the valuation rest on the right figures rather than outdated records.
Some AI tools go a step further and go through your documents for you. Document analysis can read condition reports and energy labels and pick out the factors that affect what the home is worth. In one example, it identified three critical repairs to the roof structure and sewer pipes, estimated at 108.000 kr., and a total repair budget of 218.000 kr.
For cooperative homes, a special set of requirements applies if you want individual improvements taken into account. They must be documented and approved by the board, they must be assessed by an expert, and they must be depreciated over time using the association’s method. It is this documentation that decides whether an improvement can be included and how it is weighted over the years.
Recommended price, the Marketing Practices Act and a free valuation
When you see a price in a property listing, there are rules on what the words may mean. The Danish Consumer Ombudsman’s guidelines state that the term ‘recommended price’ (vejledende pris) may only be used in a price comparison if the recommended price really is the market price. Conversely, the term ‘market price’ (markedspris) must not be used for an average price. This determines whether a listing gives a fair picture.
There are also requirements for the comparison itself. In a price comparison, the prices being compared must have been obtained at the same time; otherwise they tell you nothing useful. The marketing is misleading if the asking price has deliberately been set lower or higher than the seller will accept. A price that has been pushed up or down to provoke a particular reaction does not give you an honest picture of what the home costs.
AI valuation versus an estate agent’s valuation
When you compare an AI property valuation with an estate agent’s valuation, it is not a question of choosing one over the other. They complement each other.
The AI valuation is fast and data-driven. It builds on the fixed facts you now know about, and it can give you a well-founded benchmark straight away, without anyone having to visit your home first.
A professional estate agent’s valuation offers something else. An agent can assess condition, views and local factors that the algorithm cannot see. These are often the factors that make the biggest difference in practice, and no register captures them. That is why an AI valuation does not replace an estate agent’s valuation; it complements it.
The practical next step
At Brikk, you can get a free valuation, where you get an expert look at your home and an understanding of what the data and the market indicate, without it costing you a thing.
Leave a comment