Data Sources and Methodology Statement

Keyva's registered figures are computed from Dubai Land Department open data, obtained from Data Dubai (https://data.dubai) through its API, and from the Department's own website.

That data is the Department's records: sales and other registrations, rent contracts registered through Ejari, projects, buildings and units, and the registers of brokers, offices and valuers. The three datasets most of the figures rest on are the Dubai Land Department real estate transactions, the Dubai Land Department rent contracts and the Dubai Land Department projects.

Two tables of totals that the Dubai Statistics Center builds from the same registrations, also obtained from Data Dubai, are used where a figure's credit says so.

Where Keyva shows figures for Ajman, they come from the open data of the Department of Land and Real Estate Regulation, Ajman, which that publisher states is published under CC BY 4.0. They rest only on what its file prints: for each project, unit type, rooms count and quarter, a count of sale transactions, their total and the highest one. Where a page shows an average, it is that total divided by that count, and says so.

These are the sources of every registered figure on Keyva: not a listing site, a survey, or an estimate of our own. Asking prices are shown as asking prices, labelled as such, and never enter a typical price, a typical rent, a trend or a yield anywhere on the site.

Each figure is credited where you read it: the publisher's name sits beside it, with the date on the file it was taken from. The publishers are listed at /data-sources.

Keyva designs and controls the architecture, methodology, logic and computation behind every figure it computes. The source supplies records; Keyva decides which records count, over what period, from how many, and how they are combined.

Which records count. Prices and rents count homes only, on both the sale and the rent side; counts of plots and units are published too, where a page labels them as such. A sale counts only at a price per square foot between AED 200 and AED 20,000 and a total price of at least AED 100,000, because the register also records part payments, transfers between related parties and token amounts that are not prices. Bulk leases, where one contract covers many units at one price, are left out of rents.

Over what period. Each figure is worked out over a window it names, such as the last twelve months or the 30 days to a stated date. The newest month is always treated as incomplete, because a recent sale may not be in the file yet.

From how many. Each kind of figure has a minimum number of sales or contracts behind it, set out in the table at /how-we-compare#floors. On an area, project or building page, a figure under its minimum is shown with a red (i) that says how many records it rests on, where the page shows it at all, and left out below the smallest floors: a rent by bedroom on fewer than three contracts is not shown. A change over time, a comparison between two figures, a ranked list and the sentences written for search engines and assistants still wait for the minimum.

How they are combined. A typical price is the median: the middle of what sold. A yield is gross, before costs: a year of rents on new lets set against the sale prices of ready homes over a twelve-month window the page names. Service charges, commission and empty periods are not taken off.

So Keyva's computed figures (typical prices and rents, yields, changes, shares, and counts over a window) are Keyva's own computation from the source rows, under Keyva's architecture, methodology, logic and computation, and not a republication of the source. Where a page shows individual records as published (a building's recent sales, a registered sale on the home page, an entry in the broker register, Ajman's file totals), they are shown as published and credited to their publisher. The method is set out in full at /how-we-compare.

Keyva does not republish the source files, and does not publish an inventory of them: no count of the rows held of a dataset, and no file-by-file list of dates. Where a page lists registered sales, it shows chosen fields, not the record as filed: a building page's recent sales give the date, the rooms, whether it was off-plan, the size, the price and the price per square foot; the home page's latest sales give the area, the type of home, the size and the price.

Keyva is not affiliated with, or endorsed by, the Dubai Land Department or RERA. It does not claim a partnership with, an endorsement by, or verification by the Dubai Land Department, Data Dubai, RERA, the Department of Land and Real Estate Regulation, Ajman, or any other government body, and makes no claim that any of them has checked or approved its figures.

Keyva names no licence for the Dubai Land Department's data and claims none in its name. It names the Department as the publisher of every figure computed from that data.

Keyva's rule is that a figure names the period it covers and the homes it counts: an area's typical sale price, for example, is given for ready and off-plan homes together, over the 30 days to a stated date (or, where those days hold too few sales, the newest month with enough), from a stated number of sales.

The home page prints the date its latest records run to, and the figures are updated as the Land Department publishes.

A blog post that has been corrected says so at its end, with the date of the correction.

Questions about this statement, or a figure you think is wrong: ravi@keyva.ae.