What we find on Dubai websites

How many Dubai businesses has foundd scanned?

foundd scanned 488 businesses in Dubai between 7 June 2026 and 18 August 2026. Across the Netherlands, Romania and the United Arab Emirates together, the foundd network has scanned 1,616 businesses.

Each business is counted once, whichever scans it received and however often it was scanned, because repeat scans of the same domain collapse to the most recent one. A business scanned with three scan types appears once under each type, so the per-scan counts further down do not add up to 488.

How much work is there?

Across the 488 businesses we scanned in Dubai, our scans identified 13,343 hours of remediation work: an average of 27.3 hours per business, and a median of 21.5. Across the whole foundd network, 1,616 businesses carry 42,667 hours between them.

The gap between the average and the median is the interesting part. A minority of sites carry a disproportionate share of the total, so the average is pulled upward by them and overstates what a typical business faces. The median is the better guide to what you would find on any one site.

The biggest single group is 15 – 30 hours: 49% of the businesses we scanned. At the other end, 62 of them — 13% — need more than 50 hours.

What is an hour, and whose?

Every hour on this page is an hour of professional work — a developer, an SEO specialist or a designer, working at normal pace. It is not an estimate of how long a business owner would take doing the same job themselves. That would be longer, and we do not measure it. The median figure above, 21.5 hours, is about 2.7 working days of somebody’s time.

We publish hours rather than only scores because hours answer a question a score cannot. A score tells you a website has problems. It does not tell you whether clearing them is an afternoon or a fortnight. Hours do — and, unlike scores, hours add up, which is why this page can say something about Dubai as a whole rather than only about individual websites.

What that represents in money

foundd does not price work on this page and nothing here is a quotation. But an hour only becomes tangible once a rate is attached to it, so here is the arithmetic at prevailing Dubai rates.

Published 2026 rate guidance for the UAE puts mid-level contract web developers at AED 200–350 an hour and agencies at AED 250–700. The two bands overlap between AED 250 and AED 350, and that overlap is the range used below: the most conservative reading of both.

The typical Dubai business we scanned is carrying somewhere between AED 5,400 and AED 7,500 of unbooked web work. Across all 488, that is AED 3.3M to AED 4.7M sitting undone.

Rates from published 2026 UAE guidance (Abbacus Technologies, Cost of Hiring Web Developers in Dubai in 2026), which states it draws on local market surveys, job listings and recruitment data but publishes no sample size. We use it as an order-of-magnitude reference, not as a price list, and we show the band rather than a single rate for that reason.

How bad, and how hard?

Our scans recorded 4,910 findings across the 488 businesses, an average of 10.1 each. Of those, 2.4 per business are classified critical, and 94% of businesses carry at least one. The same scans also recorded an average of 3.4 things each site already does well.

Findings are graded into four complexity tiers, from tier 1 (an afternoon’s work) to tier 4 (structural). Splitting the 13,343 hours by tier shows where the burden actually sits:

About 28% of everything we find is a tier 1 quick fix — but quick fixes account for only 7% of the hours. The count is encouraging; the workload is not.

How do Dubai websites score?

Each scan produces a score out of 100. It is foundd's own composite measure, produced by the scan engine from a crawl of publicly accessible pages. It is consistent within the foundd network and over time, which is what makes the comparison below meaningful. It is not an external or industry-standard benchmark, and it should not be read as one.

On 4 of the 5 scan types we can compare, Dubai sits within 1 point of the network average. The exception is the AIScan, where Dubai is 1.2 points below the network.

What the averages hide

Five averages within a few points of each other suggest five similar pictures. They are not similar at all. Splitting each scan into fixed score bands shows distributions with quite different shapes behind almost the same mean.

Bands are fixed, not quartiles, so the same picture can be drawn again later and compared. n counts the scans that returned a score: sites that refused the crawler are excluded, which is why these totals are lower than the counts in the table above. Segments under 12% carry no label; hover for the count.

Not one of the 114 Dubai sites we measured with the AIScan reached 70. 19% of SiteScans did — the same businesses, the same page, a very different ceiling.

Why AI visibility is different

An AIScan identifies an average of 49.1 hours of work per business. A SiteScan on the same population identifies 15.3. The two scans surface a near-identical number of findings — 13.7 against 13.4 — so the difference is not that AI visibility is more broken. It is that the fixes are harder.

Share of each scan’s findings graded tier 4, the most structural band.

28% of what an AIScan finds is structural work, against 0% for a SiteScan. Being invisible to AI assistants is not a checklist you can clear in an afternoon.

The sites we could not read

16 of the 488 Dubai businesses — about 3% — refused our crawler outright. We record the refusal, and we exclude those sites from every average on this page, because a site we could not read is not a low score. It is an absence of a measurement.

It is worth saying what a refusal implies, though. A site that turns away an automated reader turns away every automated reader that does not hold a specific exemption, and the crawlers behind AI assistants are automated readers. Blocking is a reasonable defence against scrapers. On the current web it is also a decision about whether an assistant can describe your business when somebody asks it to.

16 Dubai sites blocked us. Whatever those sites say about themselves, an AI assistant cannot repeat it.

What the scans find most often

The share of scanned businesses in which each improvement point was recorded. These are the same categories that appear in an individual foundd report, described in the same words, so a business that has had a scan will recognise its own findings here.

What this data is not

• Not a representative sample of Dubai. These businesses either came to us or were approached by us. They were not drawn at random from the Dubai business population and nothing here is weighted to correct for that. Every claim on this page is a claim about the businesses we scanned, and it should be quoted that way.
• Not a trend. Scanning has run continuously rather than in discrete periods, so there is no earlier measurement of the same businesses to compare against. This page publishes no growth, no decline and no improvement. When we re-scan a fixed set of the same businesses, change will be reported for that set alone, like for like.
• Not an external benchmark. The score is foundd's own composite measure. It is comparable to other foundd figures and to itself over time. It is not comparable to a score produced by any other tool.
• Not a ranking of named businesses. Only aggregate figures appear here. No individual business, domain or score is identified on this page or in the dataset behind it.
• Not a price. Where the scans estimate remediation effort, that is an engine estimate of hours. It is not a quotation. The hours describe professional work at normal pace, not what an owner would take doing it themselves.
• The rates are not ours. The hourly band used to express hours in money comes from published 2026 UAE rate guidance, linked in that section. It is there to make hours tangible. It is not a foundd price and not an offer, and what the work would actually cost depends on who does it.
• Score bands are fixed, not relative. Under 40, 40–55, 55–70 and 70 and above are constants in our code. They do not move with the data, so a later snapshot can be compared with this one.

How the figures are produced

A scan is an automated crawl and analysis of a business's publicly accessible web pages by the foundd scan engine. A business may receive one or more of five scans: the SiteScan for website quality, the ShopScan for e-commerce, the SEOScan for search visibility, the AdScan for advertising readiness and the AIScan for visibility to AI assistants.

Each scan returns a score out of 100 and a set of findings grouped into categories. Businesses are deduplicated by domain and the most recent scan is kept. Scans that the target site refused are counted but excluded from every average.

A figure appears here only when its sample supports it. An average needs at least five businesses. A comparison between Dubai and the network needs at least fifteen businesses on both sides. Those thresholds are applied before the figures leave our API, so a number too thin to publish never reaches this page in the first place.

These figures were generated on 18 August 2026.

Use the data

The aggregate figures behind this page are published as a machine-readable file under a Creative Commons Attribution 4.0 licence. Reuse them, including commercially, with attribution to foundd and a link to this page.

• stats.json — the full aggregate payload, including every sample size and the thresholds applied

How to cite this page

The date on this page

The figures above are a snapshot of foundd’s scan record taken on 18 August 2026. They are not updated on a fixed schedule, so treat that date as part of the figure: anything quoted from this page should carry it.

Coverage grows as more businesses are scanned, so a later snapshot will show a larger sample. That is a change in what we have measured, not a change in the market.