The one thing every off-plan buyer wants to know, and cannot be told

Take any project in this corpus and look at what a portal actually publishes about its timing. There is a handover date. There is a construction status. There is nothing that says what date the developer announced at launch, and nothing that records the day the building was certified. The unparsed source field is not a second date hiding in plain sight either: of the 3,924 records that carry it, zero disagree with the parsed date beside them. It is the same field, written twice.

So an on-time delivery rate cannot be computed here, and the interesting part is that the same is true of every portal in this market, because they all produce the same shape of record. Any published developer punctuality score built on portal data is therefore doing one of three things: differencing two snapshots it took itself over time, which is legitimate and takes years to accumulate; buying the land registry feed,which is legitimate and costs about USD 8,170 a year; or comparing a stated handover date against today and calling the gap a delay, which is neither.

That third move is available to us. We can see 169 off-plan projects, 6% of the 2,843 that carry a date, whose stated handover has passed while the source still records them as under construction. We publish that count on the developer page, and we deliberately do not rank anyone on it. One observation cannot separate a genuinely late building from a listing nobody refreshed after it completed, and a league table would be asserting the first. It is a flag to go and check, not a finding.

382 of the 676 named developers in the corpus have no building recorded as handed over. Between them they are selling 801 live off-plan projects, 27% of the entire tracked book. Narrow it to developers substantial enough to have a scorecard at all, 3 or more tracked records, and it is still 85 developers carrying 452 live projects.

The framing matters more than the number, so here it is in full. This is an absence of evidence, not evidence of absence. The corpus sees what three consumer portals list, and it sees completions overwhelmingly from the last six years. A master developer that handed over a township in 2014 and a builder that has never completed anything both appear here as a blank, and the corpus cannot tell them apart. What the number does establish is that a buyer cannot verify a track record from any listing on any portal for more than a quarter of what is on sale, which is a fact about the market’s transparency rather than about any developer in it.

1,034 of the 1,087 completions on record, 96%, handed over in 2020 or later, and the single busiest year on record is 2025 with 328. That is not the shape of the Dubai construction cycle, it is the shape of portal coverage: a building completed in 2014 stops being listed, and a corpus built from live listings forgets it. Any statement about a developer’s history from this data is a statement about the last six years of it.

The record is concentrated as well as recent. The five largest builders account for 29% of every completion on record, the top twenty for 52%, and the top fifty for 67%. Only 78 developers clear 3 completions, which is the floor at which we are willing to describe a delivery span at all. Everyone else gets a count and no narrative.

Across the 70 developers that have both a delivery record and a live book, the median builder is selling 1.45 off-plan projects for every completion on record. The top of that table is a different business. Object 1 carries 58 live off-plan projects against 3 on record, a ratio of 19.3. Samana runs 11.2 on 56 live projects.

Read it as exposure, not as a verdict. A developer that launched in 2021 and is now selling forty projects has exactly the profile of a company scaling successfully and exactly the profile of one that has taken on more than it can build, and this ratio cannot distinguish them. What it does say is where a buyer is relying on a promise rather than on a pattern, and that is the number worth knowing before signing.

A developer average across a whole portfolio is close to meaningless: it mixes a waterfront tower with a suburban townhouse and reports the middle. So this measure holds the address fixed. Each project’s median rate per sqft is divided by the median rate of the other builders’ projects in the same community, and the developer’s own projects are excluded from every benchmark it is measured against, so a builder holding most of a masterplan is never quietly compared with itself.

Mr Eight comes out highest of the 63 developers with 5 or more comparable projects, at 1.65x, which reads as 65% above its neighbours across 8 projects. Karma is lowest at 0.68x, 32% below its neighbours. The two are 2.44x apart for floor area at the same postcodes. The median of all 1,447 project-level comparisons in the corpus is 1.00x, which is the check that the definition is not biased: a leave-one-out ratio should centre on 1 and it does.

It is a reasonable expectation that the builders charging a premium would also collect it faster, and that the cheap end would compete on generous schedules. Across the 59 developers that clear the floor on both measures, the rank correlation between how far above its neighbours a builder prices and how deep its payment plan discount runs is -0.024. That is nothing.

Split rather than correlated, it is the same answer. The 34 developers pricing at or above their neighbours carry a median plan discount of 5.73%. The 25 pricing below carry 5.79%. Those are the same number. Price and terms are two independent things a developer decides, which is exactly why both sit on the scorecard: if they moved together, one of them would be redundant, and a buyer comparing two names has to look at both.

The last measure is the only one that scores the developer rather than the buildings, and it is the only one anyone can fix in a week. Across the 111 developers with 5 or more off-plan projects, we count how many of five basic listing facts each project publishes: a price, a handover date, a unit inventory, a payment plan and a map coordinate. The market manages 86%. LEOS International publishes 100%. ANWAJ publishes 40% across 7 projects, and 7 developers sit under seventy.

Concentration is the other thing worth reading off a card. 10 of the 111 developers with a footprint measure build in exactly one community, so a buyer taking two units from them takes one bet twice. Emaar is at the other end with 26 communities across 240 live projects, a spread book. Neither is better. They are different risks, and the card says which one you are taking rather than grading it.