Two property reports can show different numbers without either being wrong. They may cover different dates, transactions, locations or measures. Before comparing their headlines, reconstruct what each number actually represents.
The objective is not to choose the more attractive result. It is to determine whether the results answer the same question.
Write down six fields for each figure
| Field | What to record |
|---|---|
| Reporting period | The transaction window, including start and end dates |
| Snapshot date | When the underlying records were retrieved |
| Geography | The exact district or project boundary |
| Transaction population | Sales, leases or another explicitly defined group |
| Measure | Count, total value, mean, median or an index |
| Unit and filters | AED, AED/sqft, property type, layout and any exclusions |
These fields are an editorial comparison method. They are not a claim that every data provider exposes the same filters. If a field is missing, record it as unknown and ask the publisher to clarify.
ADREC's market data service distinguishes sales, leases and price indices and provides filtering controls. The presence of multiple measures is a reason to preserve the selected settings whenever you cite a result.
Mean and median can differ substantially
Knownable's August archive contains a total recorded value of AED 12,329,803,236 across 2,274 sales. Dividing the total by the count produces an approximate mean price of AED 5.42 million. The separately supplied median sale price is AED 3,394,039.
Both describe this dataset, but they are different statistics. The mean divides the combined value by the number of observations. The median describes the middle of the sorted observations. A smaller number of high values can affect the mean more strongly.
This calculation does not establish a typical apartment budget. The sample combines properties and locations. The useful result is recognising that two reports using the words average price differently may be comparing different measures.
Reporting period and capture date are separate
Our edition covers August 2026 and was captured on September 6. Its dataset coverage end is September 4. That does not turn it into a report about August plus four days of September. The report period specifies the monthly window; the coverage field describes the wider dataset available at capture.
When comparing another release, distinguish its publication date from the transaction period. A report published in September could cover a month, a quarter, a year to date or a different interval.
Revisions are another possible explanation for discrepancies. A later snapshot can contain corrections or additional records. Without the earlier snapshot and an explanation of the changed records, do not claim to know exactly why a number moved.
A district name can conceal a scope difference
Marketing names, master communities, buildings and registry districts are not necessarily the same comparison unit. A report covering one development should not automatically be checked against the whole district's total.
Record the project identifiers where available. If a source does not explain its grouping, disclose that limitation rather than assigning a precise share of a wider market.
When is there a real unresolved conflict?
There is an unresolved conflict when the period, capture date, population, geography, filters and measure align but the values still differ. At that point, ask for the calculation definition and record handling rules.
Do not average two conflicting numbers to make the problem disappear. Do not choose whichever supports the planned article. If the discrepancy affects the conclusion, keep the claim unpublished until the source can explain it.
For reusable citation details, see the archived report methodology. This is general information about interpreting data, not investment advice.