Website speed in the UAE: Core Web Vitals explained, and how to fix a slow business site
By Dr. Hani Elimam Manager & CEO, SmartTech Studio
On 14 November 2026 we ran Lighthouse on our own homepage in mobile mode and got a Cumulative Layout Shift of 1.0. Google’s threshold for a good CLS is 0.1 or less, so the page was ten times over the line, on the website of a company that builds websites for a living. The cause was a single decision in our own animation script, and once we found it, the fix was small.
That is the honest starting point for this guide. Most slow UAE websites are not slow because of one dramatic failure but because of a handful of ordinary decisions: a hero image far larger than the screen showing it, an Arabic font loaded in every weight, a chat widget and tracking pixels that start before the page has painted, an animation that hides the headline until a script runs. Together they decide whether a visitor on a phone in Sharjah sees your offer straight away or stares at a half-built page.
Below we explain the three Core Web Vitals and Google’s thresholds, the difference between Lighthouse lab scores and field data from real Chrome users, what Google says about speed as a ranking factor, the fixes we would make first on a typical UAE business site, and when a rebuild is better money than patching.
The worked example: how one moving image gave our homepage a CLS of 1.0
Our homepage opens with a full-width panorama photo behind the headline, one image for each market we serve. To make the photo drift slowly sideways, our script changed its CSS left property from 50% to 0 after the page had loaded. On screen it looked like a gentle pan. To the browser it was a layout change: whenever left changes, the browser must recalculate where the element sits, and a viewport-sized element jumping from one position to another is exactly what CLS measures. Lighthouse in mobile mode scored the page a CLS of 1.0.
The fix was to keep left fixed and move the image only with transform. A transform moves an element visually, on the compositor, without changing its place in the layout, so it does not count as a layout shift and does not force the browser to recalculate the page around it. After that change, CLS went to 0 and the Lighthouse mobile performance score on the live page went from 54 to 83. The design did not change at all. The picture still moves; it simply moves the cheap way.
The same audit turned up a second problem. The largest text element in the hero, which the browser counted as the Largest Contentful Paint, was hidden at first by a fade-in animation and revealed only when our motion script started. That pre-hiding added about 1.5 seconds of render delay to LCP: the text had arrived, but the browser was not allowed to show it. The hero text now renders immediately, and the animation is kept for elements further down the page.
The lesson for almost any site: what you animate matters more than how much. Transform and opacity are cheap; left, top, width, height and margins force layout work and can shift the page. And anything that hides the main content until JavaScript runs is a direct tax on LCP, however elegant the reveal.
Core Web Vitals in plain language: three numbers and Google’s thresholds
Core Web Vitals are three measurements Google uses to describe how a page feels to a real visitor: how fast the main content appears, how quickly the page reacts when someone taps or types, and whether things jump around while it loads. Google publishes a “good” threshold for each on web.dev and recommends judging a page at the 75th percentile of visits, separately for mobile and desktop. In practice that means at least three out of four visits need to be good, not just your own visit on the office Wi-Fi.
| Metric | What it measures | Good | What usually breaks it |
|---|---|---|---|
| LCP (Largest Contentful Paint) | Time until the largest image or text block in view is rendered | 2.5 seconds or less | Oversized hero images, slow server response, render-blocking CSS and fonts, content hidden until a script runs |
| INP (Interaction to Next Paint) | How long the page takes to respond visually after a tap, click or key press, across the whole visit | 200 milliseconds or less | Heavy JavaScript, chat widgets, tag managers and pixels, long tasks on the main thread |
| CLS (Cumulative Layout Shift) | How much visible content moves unexpectedly while the page is in use | 0.1 or less | Images without width and height, late banners and embeds, web fonts swapping in, animating layout properties |
INP replaced First Input Delay as the responsiveness metric in March 2024. It watches every interaction during the visit, not just the first, so it catches pages that load quickly and then feel sluggish. On business sites, LCP is usually the hardest to get right, CLS the easiest to fix once found, and INP the one most often damaged by things added after launch.
Lab data and field data: why Lighthouse and Search Console disagree
There are two kinds of speed data, and confusing them causes most of the arguments we hear. Lab data comes from a controlled test: Lighthouse, built into Chrome and behind the lower half of PageSpeed Insights, loads your page once on a simulated mid-range phone with a throttled connection. It is repeatable enough to debug with, and it is where our CLS of 1.0 and the score of 54 came from.
Field data comes from real visitors. The Chrome UX Report (CrUX) collects anonymised timings from Chrome users who have opted in, and PageSpeed Insights shows it at the top of its report, for the URL or for the whole site, over the previous 28 days. The Core Web Vitals report in Google Search Console is built on the same kind of real-user data. It reflects the phones, networks and pages your visitors actually used.
The two often disagree, and for good reasons. Lighthouse measures one load with no clicks, so it cannot measure INP at all; it reports Total Blocking Time as a stand-in. It counts only the layout shifts during that load, while field CLS covers the whole visit, including shifts caused by scrolling and late banners. And a small business site may have no field data at all yet, because CrUX only reports pages and sites with enough traffic.
So use field data to decide whether you have a problem and whether a fix worked for real people, remembering that it trails reality by four weeks, and lab data to find the cause and check a change before you ship it. Do not chase the Lighthouse score for its own sake: it moves a few points between runs, so test several times before you trust a change.
How much does speed matter for Google rankings?
Google’s own answer, in its Search Central documentation on page experience, is measured. Core Web Vitals are used by its ranking systems, and its core systems look to reward content that offers a good page experience. But the same page says that Google always seeks to show the most relevant content, even if the page experience is sub-par, and that a great page experience can contribute to success when there is a lot of helpful content for the same query. It adds that there is no single page-experience signal, and that good results in the Core Web Vitals report do not guarantee top rankings.
In plain terms, speed is a tiebreaker between pages that answer a search equally well, not a substitute for answering it. A fast page about the wrong thing will not rank. A slow page that is clearly the best answer can still rank, but it is giving away an advantage, and on competitive UAE searches such as web design in Dubai, where dozens of pages are relevant, that advantage counts. If your site is not ranking at all, speed is rarely the first cause; we go through the more common ones in why a website does not rank on Google in Dubai, and the fixes.
The bigger payoff is usually after the click. A visitor from a paid ad, waiting on a phone for a hero image, can leave before seeing your phone number or WhatsApp button. Speed work tends to earn its cost on conversion first and search second.
Hosting, CDN and caching: the time lost before anything renders
Before anything appears, the browser has to reach your server and receive the first byte of HTML. If that server is far away, overloaded or building each page from scratch, every other metric starts late. Time to First Byte is not a Core Web Vital, but it sits inside LCP: a slow first byte leaves less of the 2.5 seconds for everything else.
For a site whose visitors are mostly in the UAE, distance matters. A server in Europe or the United States adds network round trips that a visitor in Abu Dhabi feels on every uncached request. One answer is to host in the region: the major cloud providers now run data centres in the UAE. The other, usually simpler for a business website, is a content delivery network (CDN) in front of the site, so pages, images, fonts and scripts are served from an edge location near the visitor. The large CDNs have points of presence in the UAE.
How the site is built matters as much as where it is hosted. A WordPress site with a heavy theme and many plugins builds each page on every visit unless caching is set up properly; a static site, like ours, serves ready-made files with almost nothing to compute.
Then set cache headers. Fonts, images, CSS and JavaScript files whose names change with their contents can be cached for a year with Cache-Control: public, max-age=31536000, immutable. Without headers, returning visitors download the same files every time. Missing cache headers on fonts and images is one of our most common findings on slow sites, and one of the cheapest to fix.
Arabic web fonts: the weight nobody counts
A bilingual UAE site usually loads at least one Latin and one Arabic font family, often in several weights. Arabic fonts are heavier, because each letter takes several forms depending on its position in the word, and a complete font also carries ligatures and diacritics. Four weights of a full Arabic font plus a Latin family can outweigh the rest of the page.
The fixes are well established. Serve fonts as WOFF2, the most compressed format. Subset each file to the characters you use: Arabic and its punctuation for the Arabic font, basic Latin for the Latin one. Use a variable font, where one file covers every weight, or limit yourself to two weights. Self-host the files, preload only the one or two used above the fold, and set font-display: swap so text appears at once in a fallback font instead of staying invisible.
Font swapping has a cost of its own: if the fallback has very different proportions, the text reflows when the web font arrives, and that reflow shows up as layout shift. Choose a fallback with similar metrics, or tune it with the size-adjust and ascent-override descriptors. We self-host our fonts as WOFF2, use a Latin-only subset of the headline font, and use a single Arabic family across the whole site. For the wider picture of Arabic layout, see our guide to bilingual and right-to-left web design in the UAE.
Images: usually the largest file on the page, and often the LCP
On most business websites the LCP element is an image: the hero photo, a banner or a product shot. That makes the image pipeline the single most valuable thing to get right.
Serve modern formats. WebP is supported by every current browser and is typically much smaller than JPEG at the same visual quality, and AVIF is often smaller still; our own portfolio screenshots are stored as WebP. And serve the right size: use srcset and sizes so that a phone downloads a phone-sized image, not the desktop version that is several times wider.
Then tell the browser what matters. Give the LCP image fetchpriority="high" and never lazy-load it; lazy-load images further down the page with loading="lazy". Always set width and height attributes, or a CSS aspect-ratio, on every image, video and embed. Without them the browser does not know how much space to reserve, so the text below jumps when the image arrives. Images without dimensions are one of the most common causes of layout shift we find, and the fix costs nothing.
Third-party scripts and animation: where INP problems come from
The scripts you did not write are often the heaviest on the page. A live chat widget, a Meta pixel, a TikTok pixel, a Google Ads tag, a heatmap tool and a cookie banner each download their own JavaScript and run it on the same main thread that has to respond when a visitor taps your menu. This is where most INP problems come from, and they tend to arrive after launch, one marketing request at a time.
Audit them every quarter: list every third-party script, who asked for it and whether anyone still uses its data, and remove what is unused. Load the rest after the page has rendered: we load our own analytics only once the browser is idle, and recommend the same for any tag the first view does not need. For chat, a lightweight button can load the full widget only on click; on many UAE sites a plain WhatsApp link does the job with no script at all.
Heavy animation libraries belong on the same list. A scroll-animation library that runs on every frame, across every section, keeps the main thread busy exactly when the visitor is trying to interact. Our own site uses animation, so this is not a case against it. Use it where it adds meaning, animate transform and opacity only, never hide the main content behind it, and respect the reduced-motion setting in the visitor’s operating system.
A prioritized fix list for a slow business website
If we were handed a slow UAE business site tomorrow, this is the order we would work in. The cheapest changes with the biggest effect come first, and measurement comes before everything.
| Step | Fix | Metric it helps | Effort |
|---|---|---|---|
| 1 | Run PageSpeed Insights on the home page and the top landing pages, mobile first, and note any field data | All | Small |
| 2 | Find the LCP element and make sure nothing hides or delays it: no fade-in, no lazy-loading, fetchpriority="high" on images | LCP | Small |
| 3 | Resize images and convert them to WebP or AVIF, with srcset and sizes | LCP | Small to medium |
| 4 | Add width and height (or aspect-ratio) to every image, video and embed, and reserve space for banners | CLS | Small |
| 5 | Stop animating layout properties such as left, top, width and height; use transform and opacity | CLS, INP | Small |
| 6 | Remove unused third-party scripts and defer the rest until after the page renders | INP, LCP | Small to medium |
| 7 | Subset and self-host fonts as WOFF2, limit the weights, preload the above-the-fold file | LCP, CLS | Medium |
| 8 | Add long-lived cache headers for fonts, images, CSS and JavaScript | Repeat visits | Small |
| 9 | Put a CDN in front of the site, or move hosting closer to your visitors | LCP (first byte) | Medium |
| 10 | Cut the JavaScript the site ships: drop unused plugins, split or replace heavy libraries | INP, LCP | Medium to large |
Re-measure after each step rather than shipping everything at once, so you know which change moved which number. Then give the field data time: the Search Console report reflects a fix only as the 28-day window fills with new visits.
When a rebuild beats patching
Patching works when the site is basically sound and a few specific problems drag it down, as on our own homepage: two targeted fixes, no redesign. It stops working when the problems are structural: the theme or page builder loads large amounts of CSS and JavaScript on every page; removing one plugin breaks another; the platform will not let you control image sizes or cache headers; the Arabic version is a translation plugin layered on top, with its own scripts and shifts; or every fix is undone by the next update.
Then the cost of patches keeps rising while the ceiling stays low. A rebuild on a lean stack, with images, fonts and scripts handled from day one, is often cheaper over a year than repeated rounds of optimisation, and it lets you fix content and structure at the same time. Our business websites start from AED 3,000 excl. VAT, and our guide to how much a website costs in the UAE explains what moves the price. You can see how we build new sites on our web design and development service page.
If you are not sure which side of the line your site is on, send us the address. We will tell you honestly whether it needs a few targeted fixes or a fresh start, and what each would involve.
Sources
FAQ
What are good Core Web Vitals scores?
Google’s published thresholds on web.dev are an LCP of 2.5 seconds or less, an INP of 200 milliseconds or less and a CLS of 0.1 or less. A page passes when all three are good at the 75th percentile of real visits, measured separately for mobile and desktop.
Will a faster website rank higher on Google?
It can help, but it will not override relevance. Google says Core Web Vitals are used by its ranking systems, yet it always seeks to show the most relevant content even when page experience is sub-par, and good Core Web Vitals results do not guarantee top rankings. Speed matters most when many pages answer a search equally well.
Why does PageSpeed Insights show different numbers from Search Console?
The lower half of PageSpeed Insights is a Lighthouse lab test: one load on a simulated phone, which varies from run to run and cannot measure INP. Search Console shows field data from real Chrome users over the previous 28 days. Use the lab test to find causes and the field data to judge results.
Does hosting my website in the UAE make it faster?
For visitors in the UAE, a nearby server or a CDN with an edge location in the country shortens the time to the first byte, which feeds into LCP. A CDN in front of the site is usually the simpler option. It will not fix heavy images, fonts or scripts, which cause most slow pages.
Should I speed up my current website or build a new one?
Fix the current site if a few specific problems slow it down, such as oversized images, a layout shift or late scripts. Rebuild if the theme, plugins or platform are the bottleneck and fixes keep being undone. SmartTech Studio builds business websites from AED 3,000 excl. VAT.