Data & Methodology

How Do We Calculate Value?

A number is only as good as what it’s based on. Here we clearly explain what data Instant Valuation uses, how it selects comparables, how it validates the result, and where it might be wrong.

  1. DataInformation from the form, location, current listings in the area and our own listing history
  2. ComparableThe most similar listings based on distance, size and recency
  3. EstimationThe area’s price per m² band and the property’s characteristics
  4. ValidationConsistency checks and confidence level

1. What data do we use?

  • Information you provide: property type, square meters, rooms, floor, building age, aspect, parking, complex and other amenities; the location you marked on the map. The marked point is compared with the selected district; if too far, the district centre is used.
  • Current listings in the area: artificial intelligence scans similar listings in your property’s area from listing sites over the last three months and derives a price per m² band from them.
  • Our own listing history: similar listings we have published in the last 24 months — both active and expired. This data serves as an anchor that validates our analysis; when artificial intelligence cannot be used, the value is calculated directly from these listings.

2. How Do We Select Comparables?

We assign a similarity score to each listing; the lower the score, the more similar the listing. The score is the sum of three components:

  • Distance: 1,5 km equals one point. If there is no location, a listing in the same district receives one point; listings in other districts receive a higher score.
  • Size: A square metre difference of around %25 counts as one point. The difference is measured proportionally; 100–125 m² is considered the same as 200–250 m².
  • Recency: One point is deducted for each year the listing is outdated.

Outlier values incorrectly entered in the price per m² of the 6 most similar listings are statistically removed (using the interquartile range rule) and the median is used instead of the average. This median forms the basis of the valuation; the same listings are shown in the report as "comparables used for valuation".

3. How do we estimate the value?

The valuation is based on the area’s price per m² band multiplied by your property’s square meters. Floor level, building age, aspect, being in a complex and similar features determine where your property sits within that band; these are listed in the report as “value-increasing” and “value-decreasing” factors.

If artificial intelligence is unavailable at that moment, the value is calculated by multiplying the median price per m² of the most similar listings by the square metres of your property. In this case, features such as floor level, age and aspect are not taken into account and the confidence level is at most "medium". For a "medium" rating, the comparables must be in a narrow price range and close to your property (including marking the location on the map for sales). If comparables are in other districts, the district's own current average listing price is preferred; if there is no data in the area, the province's average price per m² is used and the confidence is shown as "low".

We do not take your expected price into account. We do this deliberately: when that figure is entered into the calculation, the result is pulled towards it. We make the comparison in the report after the valuation is produced.

4. How do we validate the result?

Every result generated by artificial intelligence is checked by code before publication:

  • If the highest value is more than 3 times the lowest, the result is considered meaningless and is rejected.
  • If the unit value is more than 3 times our median comparable listing or less than 3 of it (for example, using a rental figure instead of a sales price), the result is rejected. Deviations exceeding 1,5 times reduce confidence.
  • In sales valuations, the estimated rental yield is not shown in the report if the annual gross return falls outside %1–%15.

If a result is rejected, the next method is tried: first AI without search, then direct calculation from similar listings, then the area average. If none are available, your request is forwarded to our consultant and we will get back to you. We prefer not to show a figure at all rather than display an inaccurate one.

5. What Does the Confidence Level Mean?

Artificial intelligence also states its own confidence level, but this is only an upper limit: the checks above can lower it but never raise it.

  • High: 10 or more current comparables (preferably at neighborhood level), a narrow range and strong alignment with our own data.
  • Medium: 5–9 comparables, data only at city level, a range where the highest value exceeds the lowest by 1,5 times, significant deviation from our own data or inability to conduct a current listing survey at that time.
  • Low: Fewer than 5 comparables or a range where the highest value exceeds the lowest by more than double. In this case, treat the figure as a starting point.

6. How accurate is it?

We test the method with our own data: we estimate the price of each listing using the same method from all other listings (excluding that one) and compare it with the actual listing price. In our November 2026 measurement (Konya), when the location was marked on the map, half of the estimates for homes for sale were within 15% of the listing price, and within 8% for rental homes. When the location was not marked, these figures were 19% and 11% respectively. For comparison: using only the official provincial average price per m² gives a 28% error for sales.

That's why we provide a range rather than a single figure and clearly state that no automated valuation can match the certainty of a professional appraisal. An assessment by a consultant who views your property in person will narrow this range.

7. Where can we go wrong?

  • List prices are not selling prices; completed sales are usually lower due to negotiation.
  • We have not inspected your property in person. We only know about factors such as renovation, views, floor plans and building quality to the extent you have described them.
  • If there are few listings in the area, the price range widens and confidence decreases; we clearly state this in the report.
  • The result is a preliminary assessment; it does not replace an official valuation report and cannot be used for credit, collateral or legal purposes.

We use essential cookies for the site to function. We would also like to use analytics cookies to understand how visits are going and marketing cookies to show ads on blog pages; these will only work if you consent. Learn More