The State of Sameness: Dubai Real Estate, Measured

In 2021, a major Dubai property brokerage touched 30 pages on its website. In 2026 it touched 13,986. That one ramp shows what AI did to content production, and it opens this series: we point our instruments at one category at a time and measure the AI content sameness nobody can see from inside.

First up: Dubai real estate, one of the loudest content categories on earth.

The production arms race

The AI era did not create this brokerage's content strategy. It multiplied it by 80 in two years. Of the 31,725 indexed pages, roughly 98% are templated inventory: listings, programmatic buy-guides, area pages generated by pattern. About 700 are editorial articles, the content meant to carry the brand's voice.

Every brokerage in the category is running some version of this playbook, with the same AI, on the same topics. That is the setup for convergence, and convergence is what the choosing machine cannot work with.

What the machine reads

We read 136 of the editorial articles with the sameness instruments. The Brand Voice Score came back at 22.1 out of 100: some voice, and a lot of templated writing around it.

The detail that matters: one sharp explainer in the corpus proved a real voice exists inside this company. Someone there can write a piece that teaches, takes a position, and sounds like a person. It was the outlier, not the norm. The body of work averaged that voice away, which is exactly what the logo test detects: distinctiveness that exists in places and disappears in the aggregate.

The away game

The Third-Party Trust Score came back at 31.4 out of 100, and this is where the category study gets uncomfortable.

The brokerage shows up on 12 of the 21 third-party surfaces we checked, one more than the category leader, and its press footprint runs at twice the leader's. On paper, it is ahead. But its verified presence on Reddit, the single most-cited source across AI engines, is effectively zero, and roughly 85% of what AI assistants say about brands comes from third-party sources¹, with community surfaces weighted heaviest².

So the paradox: it is winning the surfaces AI skims and absent from the surfaces AI trusts. The 31,725-page operation lives almost entirely on its own domain, the one place the machine weighs least.

The verdict: the AI Commodity Trap

High owned volume. Converging voice. Thin earned trust. That combination is the AI Commodity Trap: the machine retrieves the brand constantly and chooses it almost never, because nothing in the corpus tells it this brand is not the category.

Producing more content cannot escape the trap, because more AI-default production deepens the sameness and adds it to the wrong battlefield. The way out runs in a different direction: fewer, sharper positions; proof only this brand can make; and a deliberate footprint on the surfaces the machine trusts.

What any category should take from this

1. Check your ramp. If your production multiplied since 2024, so did your competitors', with the same models. Volume is now the sameness accelerant.
2. Read your own corpus the way the machine does. Not your best piece; the aggregate. One great outlier does not survive 700 templated siblings.
3. Count the away game. Your presence where AI actually forms opinions: communities, reviews, press. That count predicts your visibility better than your page total.
4. Then hold it. The pull toward the category average never stops. A one-time fix decays; a watched position compounds.

Method note

Three deterministic modules produced these numbers: a production census, a sameness scan of the editorial corpus, and a third-party footprint check across 21 surfaces. Numbers are reported only where verified by the instrument run; brands are anonymised; no AI-engine outputs were fabricated or simulated. This is the same instrument that powers the Brand Distinctiveness Index, built by Ivanooo as part of Distinctiveness Engineering.