What Dubai Land Department sales data covers, and what it leaves out
The fields a sale record carries, the ones it does not, and where the register is knowingly incomplete.
Every figure on this site comes from sales registered with Dubai Land Department. The register is narrower than it looks, and most mistakes with this data come from misjudging its edges.
What one record contains
A registered sale is a short document. It says when the transaction was recorded, where the property sits, what kind of property it was, roughly how large, and what was paid. The room count is there when the filing included it.
That is enough to answer a lot of questions about price and almost nothing about quality. Two apartments in the same tower, one on the fortieth floor facing the water and one on the third facing a car park, produce records that differ only in size and price. Whatever explains the difference between them is not in the file.
This matters most when someone compares a specific unit against a building median. The median is real, and the unit's position relative to it is unexplained by anything you can look up.
Four kinds of property in one table
| Record type | Share of sampled sales | Sampled rows | Median AED/sqm |
|---|---|---|---|
| Unit | 81.1% | 154,043 | 18,151 AED/sqm |
| Villa | 9.7% | 18,462 | 15,112 AED/sqm |
| Land | 8.9% | 16,812 | 15,166 AED/sqm |
| Building | 0.3% | 549 | 14,213 AED/sqm |
Apartments dominate the count. Land and villas together make up most of the rest, and a few hundred sales are whole buildings changing hands.
Mixing them is the single most common way to produce a wrong number from this data. Land sells by the square metre in bulk; apartments sell by the square metre in units of eighty. Averaging the two gives a figure that no buyer of either would recognise. Every median on this site restricts itself to one record type for that reason, and where a comparison spans two, both sides are restricted the same way.
The same trap catches area pages. An area that contains a large plot sale looks different from the same area counted on apartments alone, and the difference has nothing to do with apartment prices moving.
Most of what is recorded has not been built
62%
Off-plan share of sampled sales
189,866
Sampled sales
2023-08-14 to 2026-08-13
Sample window
1,339,968
Sales in the full register
A clear majority of recorded sales are off-plan, which means the price attached to them is a launch price for something that does not exist yet. Those sales enter the register in blocks when a project releases, at prices the developer set.
Read that alongside any area median and it changes what the median means. In an area with heavy off-plan activity, the median is closer to a developer's price list than to a market clearing price. In an established area where most sales are resales, it is closer to what buyers and sellers actually agreed.
Both are accurate answers to different questions, and the register will only tell you which one you have if you split the rows yourself.
What the register does not know
| Not in the data | What that means for a figure you read here |
|---|---|
| Rent | Not in this table. Lease contracts are registered separately with Ejari and join to this data only at the area, which is what the rental yield guide is built on. |
| Service charges | Set per building and billed annually, never part of a sale record. Two towers at the same price per square metre can differ by thousands a year in running cost. |
| Floor, view, and finish | A high floor with an open view and a low floor facing a wall register identically. Most of the price spread inside a single building comes from attributes the record does not carry. |
| Buyer nationality | Published by DLD in aggregate elsewhere, not attached to individual sales here. |
| Developer | Project names often imply one, but the field itself is absent, so any developer ranking would be string-matching rather than counting. |
Rental yield is the one worth spelling out. Any yield figure needs a rent and a price, and this table holds only the price. Dubai does register leases, separately, through Ejari, and the two join at the area rather than at the building. Anyone publishing a Dubai yield without saying which rent went into it has supplied it from somewhere, and the quality of the yield is the quality of that other source.
Where the record is incomplete
Some fields are simply blank. Around 14% of sampled rows carry no room count, 35,823 carry no building name, and 16,878 carry no project name. A filter that assumes those fields exist drops every one of those rows in silence, and its count will disagree with the count on the page beside it.
There is a second, softer version of the same problem at the end of every snapshot. Sales are recorded after they complete, so the most recent weeks are always thinner than they will eventually be. The last month in any extract usually looks like a slowdown it will grow out of.
Reading a figure on this site
Check the record type first. If a comparison puts an apartment figure next to an area figure, confirm the area figure was restricted to apartments, because the unrestricted version includes villas and land and will sit lower.
Check the sample size second. A median over thirty sales moves on a handful of them. Every table here prints the count behind the number for that reason, and where the count is small the number shows a direction and little more.
Check the window third. Figures here cover 2023-08-14 to 2026-08-13. The full register runs back to 1990-02-12 and holds 1,339,968 sales worth 3.7T, but the medians published across this site are computed on the recent window, not on all of it.
Related reading: what a building median does not tell you shows how wide the spread inside a single address can be, and does the building price above its neighbourhood works through a comparison that only holds when both sides are restricted to the same record type.
Data comes from Dubai Land Department historical sales records through 2026-08-13. It covers sales only (not rentals) and is market reference, not valuation or investment advice.