Portal audit: Bayut, Property Finder, Dubizzle

Where the searches are

People search for names, not topics. A community or building name gets 10 to 1,000 times the searches of a question such as "rental yield dubai". Figures are Google Keyword Planner averages per month, UAE, all languages, September 2025 to August 2026.

Year on year, Keyword Planner shows generic listing terms down 18% to 71% (apartments for sale in dubai −56%, off plan projects dubai −77%). Named buildings and data terms go up: binghatti elite +52%, burj vista +52%, emaar beachfront +83%, dld transactions +49%, service charge index dubai +100%.

Who wins in Google

In 16 of 28 live searches no portal is in the top 3. The portals win buyer listing searches and lose tools, seller, family, yield and service-charge searches. Sample: Google, Dubai, English, desktop, one run each on 7 October 2026.

• Search-results pages do the ranking. They are 36 of the 48 portal results (Bayut 11 of 16, Property Finder 21 of 28, Dubizzle 4 of 4).
• Their content pages do not rank. No portal building page, school guide, floor-plan page, agent directory or calculator is in any of the 28 result sets. This sample is small: the traffic data in the next section shows that Bayut's building and area guides do rank, for thousands of names.
• Who takes the open ground. Brokerages (bhomes.com in 5 searches, engelvoelkers.com in 4), Reddit (4, all family and lifestyle), dubailand.gov.ae (3), dxbinteract.com (3, position 1 for "dubai property prices"), trpe.ae (3).
• No portal anywhere for "golf views seven city", "best areas for families in dubai", "schools near arabian ranches", "living in dubai hills estate", "how to sell property in dubai", "property valuation dubai", "dubai property transactions", "dubai marina rental yield", "service charges dubai marina" and all 3 calculator searches.
• AI Overview is on 24 of 28 searches. A page must hold dated, sourced numbers in its HTML to be the cited source.

Size, third-party estimates: Similarweb 3-month visits are Property Finder 13.2M (July to September 2026), Bayut 7.6M (June to August 2026), Dubizzle 16.6M across all its classifieds. Sources: Bayut, Property Finder, Dubizzle.

Where portal traffic comes from (UAE)

Search-results pages earn most of the organic traffic, and the top keywords are building and landmark names. Figures are DataForSEO estimates for Google UAE, English, from the top 1,000 pages of each domain.

• Building guides are Bayut's largest traffic source. 364 building guides in its top 1,000 pages bring about 709,000 visits a month, more than its search-results pages. The top one is /buildings/mai-tower/ for "mai tower" (33,100 searches, position 1).
• This corrects the section above. The 28-search sample found no building guide in the results; the sample was too small. Building-name demand is real, and Bayut holds much of it with editorial pages whose price tables are blank.
• Property Finder holds building names with search-results pages: position 1 for "silicon gates 1" (27,100) and "alma arabian ranches" (40,500). Its own building and area guides earn 0.6%.
• Sold-price pages earn little. Bayut's transaction pages take 2.6%; Property Finder's are not in its top 1,000. The price index pages are not in Bayut's top 1,000.
• Property Finder's blog earns 22% from place pages such as "karama shopping complex", not from property advice.

Bayut by page category: pages × visits per page

Building guides are Bayut's largest category: 1,884 guides earn about 1,045,000 search visits a month, 555 each. This table covers Bayut's top 8,000 pages in Google UAE, which hold 99% of its estimated 3.07 million monthly search visits.

• Guides earn the traffic. Building and area guides are about 10,600 pages, 2% of what Bayut publishes, and bring 53% of its search visits.
• Rent beats sale. To-rent search pages earn 3.5 times the visits of for-sale search pages.
• The data pages earn almost nothing from search. 118,870 price index and trend pages bring about 1,000 visits. Transaction pages bring 93,000, mostly rent transactions.
• A quarter of the building guides rank. 1,884 of about 7,358. The top one, Mai Tower, earns about 16,000 visits.
• These are search visits only (DataForSEO estimates). Brand and direct traffic, the app and paid traffic are not in the table. "Pages published" comes from Bayut's sitemaps read on 7 October; the listing files were not counted in full.

Propsearch: the site to copy

Propsearch earns about 2.18 million search visits a month in the UAE from building pages alone, with a team of 2 to 10 people and 2,373 linking domains. That is close to Property Finder (2.31 million) and it proves a small data site can rank first for building names.

Its weak points, which our page can beat:

• No rents, yields, price trends or charts on the free page; mortgage registrations are mixed into "sale transactions".
• A price display bug: AED 990,000 shows as "99k".
• No structured data, the same meta description on every page, a wrong og:url, no sitemap, English only.
• No author, no method, no updated date on building pages.
• No unit numbers: the land register's open data does not publish them.

Not found: licence type, funding, ownership, any paid price, traffic split by rent and sale intent.

This changes one earlier conclusion. The spec said head building names would stay with Bayut for 12 months or more for lack of links. Propsearch ranks first for names such as "skycourts" (49,500 searches) and "sulafa tower" (27,100) with one eighth of Bayut's linking domains. A facts-and-data page per building, built over years, is enough.

The UK and the UAE are different markets

The UAE market searches for a place to live now; the UK market researches whether to buy. The portals are almost absent in the UK: under 16,000 organic visits a month each, against 2 to 3 million in the UAE.

• UAE: building and community names, rent searches, calculators and admin. Build items 1 to 6 serve this market.
• UK: city-level terms (prices, for sale, moving, visa, cost of living). Building names and calculators get almost no UK searches.
• Area guides matter in the UK. They are 27.5% of Bayut's UK traffic and 5.2% more comes from building guides. In the UAE, "X area guide" gets 10 searches.
• A UK page set is small: Dubai property prices, house prices by area, buying from the UK, golden visa, cost of living, moving. About 10 pages, fed by the same data.

AI visibility

In AI answers about Dubai property, a brokerage and YouTube are cited more than the portals. Source: DataForSEO's database of Google AI answers, per country. It counts the answers that cite each domain.

Who is cited when the question contains "dubai property" or "dubai real estate":

• For "dubai rent" questions in the UAE the portals lead: Property Finder 37, Bayut 30, Dubizzle 26 of 510 answers.
• In the UK, wise.com, zoopla.co.uk and theguardian.com are cited for "dubai property"; no UAE data site is.
• ChatGPT data exists for the US only: 20 answers on these topics, and no portal is cited in any. Sources are brokerages, Reuters, Khaleej Times and globalpropertyguide.com.
• dxbinteract.com ranks first in Google for "dubai property prices" but is cited in only 97 AI answers. Data in an app is not quotable; data in page text is.
• Next step for a ranking that we can track: a fixed panel of 40 prompts per market, asked each week of ChatGPT, Gemini and Perplexity.

Technical crawl, speed and links

The portals are clean on the basics and slow on mobile. Sources: LibreCrawl (Property Finder, 618 URLs), a small crawler with the same fields (Bayut, 200 URLs; Bayut blocked LibreCrawl with a 503 security check), Firecrawl (Dubizzle, 12 pages; it blocked LibreCrawl too), Lighthouse through DataForSEO, and DataForSEO backlink data.

Mobile Lighthouse, one run each:

• Bayut's search and transaction pages and the Dubizzle page refused the Lighthouse run, so they have no score.
• Correction: Bayut's building guide is not blank. Seen in a browser, the Marina Gate 1 guide has 3,277 words and 4 transaction tables (recent rental and sale transactions), rental and sales trends and a rental yield section. They load by script, so the server HTML and my first reading showed blank tables. Our edge is not "data against no data". It is unit, floor and view detail, asking against sold, a stated method and author, data in the HTML, and speed.
• Bayut's transaction table in a browser shows date, building, price, bedrooms, built-up area and "vacant" or "rented at time of sale". It shows no unit number and no floor.
• Bayut opens a pop-up on its search page ("TruBroker Stories") and a prompt on its transactions page.

What the three have in common

All three are listing marketplaces paid by agents. Each page type exists to make a search page rank and to send a lead to an agent.

Popular means repeated on every page: Call, WhatsApp and Email buttons, the verified badge, building chips, "similar properties", the mortgage estimate and a templated FAQ.

What is useful, by persona

The buyer is well served. The seller is served worst, and the family and the investor get parts of an answer on different pages.

Content strategy of each portal

Bayut: the widest set of page families on one location tree. The same path, for example /dubai/dubai-marina/marina-gate/marina-gate-1/, sits under for-sale, to-rent, floorplans, brokers, companies, index, trends and transactions.

• Page families in the sitemap: price index 97,120 (sale and rent), search 87,628, sale transactions 44,138, floor plans 37,990, broker profiles 26,129, long-tail 23,549, building guides about 7,358, area guides about 3,274, school guides 1,208.
• Data is written into FAQ text. The Dubai Marina search page has 21 templated FAQs, with average asking AED 4,011,226 and average sold AED 3,279,402 in separate answers.
• Editorial runs on WordPress: "X vs Y: where should you live", "pros and cons of living in", legal how-tos, and half-year market reports for 4 emirates.
• Brand devices carry the trust: TruCheck, TruBroker, TruEstimate, Dubai Transactions "sourced directly from DLD".
• Weak point: guide pages load their numbers by script, so the HTML has blank rent, sale and yield tables. Templates slip: the Marina Gate 1 FAQ names "Masakin Al Furjan".

Property Finder: fewer families, more data on each page. Search pages are the only family in its sitemap; all else relies on internal links.

• Each search page carries a Scout AI summary, about 100 building chips, a 900-word templated essay with FAQ and a "Price insights" table that links to building transaction pages.
• The area guide is the best single page type in the sample (24 of 30): essay, market statistics, schools table, transport, resident reviews, FAQ.
• Transaction pages use a question title, "Buying Property in Dubai Marina? Know the Market First", with live numbers in the meta description.
• The blog chases high-volume UAE life-admin searches (visas, VAT, vehicle tests). Market Watch reports are PDF only; the newest is dated 3 February 2026.
• Weak point: numbers disagree between pages. Dubai Marina price per sq ft is 2,146, 2,161 or 2,749 depending on the module.

Dubizzle: classifieds reach, no data layer. It is in the same group as Bayut and borrows TruEstimate from it.

• Its own ground is rooms, bed spaces (from AED 799 a month), short-term lets and private-owner ads, with chat as the contact route.
• Search and long-tail pages follow the Bayut pattern; a "building page" is only a filtered result list.
• The blog has 7 categories and 2 to 4 posts a week, and links to Bayut for school and mall guides.
• Weak point: no sold prices, trends, schools or maps on any page. Page 1 of New Projects is all paid placement.

Gaps that none of the three fills

Each gap below was looked for on the sampled pages of all three portals and was not found.

Priorities: build, copy, skip

Build in the order below. The order is search demand, times the size of the gap, times how much of the data we hold today.

Build

Copy

• One location tree for every page family (Bayut). A building has one path; prices, floor plans, schools and yield hang from it.
• FAQ text filled from live numbers (Bayut, 21 per page). Add what they omit: the asking-to-sold gap, the date and the source.
• Chips with counts and dense sibling links (all three): buildings in a community, bedroom variants, nearby communities.
• "Similar sold transactions" and "costs X% more than the average" (Property Finder listing). Put both on the building page.
• Index table by bedroom and by building with 3, 6 and 12 month change (Bayut).
• Schools table with curriculum, fee range and KHDA rating (Property Finder), with drive time in place of distance.
• Question title with live numbers in the meta description (Property Finder transactions page).
• A visible check date on each number (Bayut TruCheck). Their index pages have no method; ours must.
• Alerts for a new sale or price change in a building (Bayut and Property Finder both offer transaction or price alerts).

Skip

• Listing search pages. The portals hold 25,000 to 88,000 each and all the inventory.
• Agent directory, badges and lead routing.
• Lifestyle, place and life-admin blog posts. They bring Property Finder 22% of its traffic but no property intent.
• AI chat. Both portal versions are beta, and Scout's yield figure contradicts its own site.
• Rooms, short-term lets and an app.

Scorecard: 49 pages, categorised and scored

The best page in the sample is the Property Finder area guide (24 of 30); the worst are the seller forms (4 and 6). Each page has six quality scores from 0 to 5 (SEO, content, data, tools, UX, trust; total of 30) and four persona scores from 0 to 5. BY = Bayut, PF = Property Finder, DZ = Dubizzle.

Page types that one portal has and the others lack: floor-plan pages and school pages (Bayut only), a standalone mortgage calculator (Property Finder only), rooms and bed spaces (Dubizzle only).

Limits, open decisions, sources

Limits of this audit

• Scores come from one sampled page per type, read as text. Only 4 pages were then checked in a browser, so UX scores judge structure and gating, not visual design. Bayut data scores in the scorecard are too low where tables load by script.
• Bayut blocked the scraper on its app pages. Those pages were read as server HTML, so modules that load by script (transaction rows, charts, agent cards, map) were not seen. Bayut data scores can be too low.
• Traffic figures are DataForSEO estimates from the top 1,000 pages of each domain (64% to 74% of UAE traffic). They are models, not analytics. Ahrefs returned "Insufficient plan".
• Markets covered are the UAE and the UK. India is not counted. The AI-answer counts cover Google AI answers; ChatGPT data is US only.
• Search results are one desktop run per query.

Decisions that are yours

1. Publishing portal-sourced data. Items 1 to 3 use unit numbers and listings collected from public portal pages. The DLD open data is free to publish but has no unit numbers.
2. The data.dubai API key. It unblocks service charges, the projects register and rent contracts (items 5 and 8).
3. The weekly AI prompt panel: 40 prompts per market on ChatGPT, Gemini and Perplexity. DataForSEO spend so far is about $4 of $51.

Sources and files

• 54 portal pages saved as markdown (49 scored), the 28 search results and the Keyword Planner volumes: held on file.
• Pricing read from DataForSEO Labs and DataForSEO Google Ads.
• Bayut product dates: About Bayut.