How We Clean DLD Transaction Data
Every figure on KhaleejKeys traces back to the Dubai Land Department's public registry. Here is how we turn the raw export into ready-market residential sales - and what we deliberately refuse to claim.
Search TransactionsThe source
Our data comes from the Dubai Land Department's official open-data transaction export - the same public ledger anyone can download. It contains every registered real-estate transaction in Dubai: sales, mortgages, and gifts, across apartments, villas, offices, shops, land plots, and whole buildings.
That raw file is not directly usable as residential market intelligence. It carries heavy duplication (each transaction appears multiple times), commercial and industrial property, financing events that aren't real sales, data-entry errors, and inconsistent area naming. Our pipeline turns it into what KhaleejKeys actually shows: completed, arm's-length, ready-market residential sales.
Where the data can't support a claim, we would rather show nothing than show something fabricated. That principle is the product.
Only genuine ready-market sales qualify
The DLD groups every transaction as a Sale, a Mortgage, or a Gift. We keep only Sales - and only ready ones.
A financed purchase appears in DLD records as two rows - a Sale and a separate Mortgage registration. We keep the sale, not the financing event. This is also why per-sale financing status is not derivable from the public data, so we never display it.
Gift transfers - often family transfers at nominal prices - are their own group and are excluded entirely.
Every sale-price figure, average, search result and trend on KhaleejKeys is ready-market resale only - off-plan is excluded from all of it. Off-plan (developer launch) sales are stored in a separate lane and surfaced only in the "Off-plan vs ready" comparator on community pages, clearly labelled - never mixed into the resale prices.
These are financing transfers, not arm's-length price discovery.
Only residential units qualify
We then reject anything that isn't a per-unit residential property.
A building changing hands is not a unit price.
Outside the residential market by definition.
Identified from the unit's own classification and removed - these aren't part of the residential market.
Sanity bands drop data-entry errors
Even genuine residential sales sometimes carry impossible numbers. Three bands apply.
Must be a plausible unit size; impossible values are dropped.
Must be positive.
This is the key error filter. The range is set wide enough to retain Dubai's genuine ultra-prime sales while removing the size or price entry errors that would otherwise wreck averages, along with nominal-value rows.
Must fall between 1 January 2000 and the import date. The raw file contains a small number of rows with implausible or legacy dates; pre-2000 records are sparse legacy entries outside the product's scope.
Deduplication
Every transaction carries a unique DLD transaction ID; we de-duplicate on that ID during import, and the database enforces uniqueness on it permanently. Re-importing the same file - or an updated file - can never create duplicates. Only genuinely new transactions are ever added.
How we categorise what's left
Three derived fields, each built from the DLD source columns.
Derived from the DLD source fields and normalised to Apartment, Villa, or Townhouse.
Normalised from DLD's room descriptions (Studio, 1, 2, …).
Community (area) name - our most careful step
DLD records locations as cadastral zone names that don't match how anyone searches for property. We resolve each one to a consistent market community name through a layered, manually-verified mapping, with the original source values preserved. Anything we can't confidently place is passed through under its raw name and flagged for review - never silently guessed. That review queue currently stands at zero.
Nothing is overwritten: full provenance
For every transaction we store both the raw DLD values (original cadastral area name, original master project) and the derived display values - plus a record of which mapping rule produced the community name and whether it was flagged for review.
Any displayed community name can be traced back to the exact DLD source values and the rule that mapped it.
Mapping improvements are applied by re-deriving from the stored raw values - the underlying record never changes.
What this produces
Applying the steps above yields the current KhaleejKeys dataset:
All property sizes are shown in square feet (sqft), automatically converted from square metres where the DLD source records area in metric.
Reading price trajectory
Alongside the raw figures we show a price trajectory for each community - a plain-language read of whether registered prices have been rising, holding steady, or cooling over the past year. It is deliberately conservative: the direction is weighted by how many sales support it, cross-checked between the most recent quarter and the full year, and adjusted for shifts in the mix of homes sold - so a year that simply saw more large or premium units change hands is not mistaken for genuine appreciation. Each read carries a confidence level, and where too few sales have been registered we show no trajectory at all rather than a figure a handful of sales can't support.
What we deliberately do not claim
Where the data can't support a claim, we would rather show nothing than show something fabricated.
Source & licensing
Property transaction records are published by the Dubai Land Department (DLD) as Open Data on the Dubai Government data portal under the Dubai Open Data Licence (Dubai Data Law - Law 26 of 2015 & Law 2 of 2016). Source dataset: data.dubai - DLD Real Estate Transactions.
Every figure on KhaleejKeys is independently processed and derived from that source by the steps above. We publish a transformed, value-added view - not a copy of the raw register.
KhaleejKeys is an independent product and is not affiliated with, authorised by, or endorsed by the Dubai Land Department or the Government of Dubai. Data is provided "as is", without warranty - always verify against the official register before relying on any figure.
Have a question?
More on what's included, how prices work, and using the data - all answered in one place.
Read the FAQ