Skip to main content
Rasad

Blog

Blog Overview

Core Web Vitals for Arabic websites: where the performance budget goes

PublisherAbwab Admin
Published OnNov 15, 2026
Reading Duration7 min read
Last UpdateNov 15, 2026
Core Web Vitals for Arabic websites: where the performance budget goes

The Arabic font, images and scripts consume the performance budget first. This article explains the LCP, INP and CLS thresholds, how to measure them at the 75th percentile, and how to fix them in order.

An Arabic website does not become fast by translating its text. The Arabic script needs a different font file, direction reorders the page, and translated copy may be longer or shorter than the original, changing spacing and line count. When the page opens on a mid-range phone on a mobile network, those differences show up in Core Web Vitals before they show up in any marketing report.

This article separates where the performance budget goes on an Arabic site: the font, the hero image, third-party scripts, hydration, and visual shift. It explains how to measure each metric at the 75th percentile of real visits, not on your fast laptop. The figures here are not theoretical targets; they are the thresholds published by web.dev.

What the thresholds actually mean

LCP measures when the largest visible element appears. Not when HTML arrives, not when loading finishes, but when the user sees the main element. INP measures the page's response to an important interaction: a click, a tap, typing. CLS measures how much elements move while loading. All three are measured from real user data at the 75th percentile, meaning 75% of visits must fall inside the threshold, not the average.

This matters because the average hides the problem. If half the visits are fast and half are slow, the average may look acceptable while half the users struggle. Watch the 75th percentile per page, and separate Arabic from English because font and content differ. A fast English page does not mean its Arabic version is fast.

Where the milliseconds go

SourceEffectWhat to fix first
Arabic fontA larger file, multiple weights, delayed text renderingTwo weights only, preload the base font, an appropriate font-display
Hero imageThe largest element on screen, often heavyModern format, correct dimensions, priority loading for that image only
Third-party scriptsAnalytics, chat and experiments block interactionDefer what is not essential, and review every tag before adding it
HydrationThe page appears, then freezes before it respondsLess client JavaScript, and split what can be split
RTL shiftElements move after the font or image loadsReserve space up front, and fix image and icon dimensions
ServerLong response time before the first byteCaching, faster queries, hosting close to the audience

The Arabic font is the first decision

An Arabic font is not ordinary text. Letters join, shapes change by position, and the file needs wider coverage than a Latin font. If you load seven weights because the design shows them, you pay on every visit. Start with two: regular and bold. Use a family such as Noto Kufi Arabic or IBM Plex Sans Arabic, and test how numbers and mixed Latin text render inside it.

Do not let the font block the whole page from rendering. Preload the base weight, and use font-display: swap or a considered fallback so text appears in the system font and then swaps. But watch out: the swap itself causes shift if the two fonts have different metrics. Test the difference between the system font and the final font on a phone screen, because CLS does not forgive.

CLS in the opposite direction

In RTL, shift is not limited to images. A sidebar may move from right to left after loading, a dropdown may open in the wrong direction, and a logo may push the heading. Every element without reserved space moves its neighbours. The fix is not cosmetic: explicit image dimensions, reserved line height, fixed space for the bar, and no ad or banner inserted above content after render.

From reading to shipping

Want this built, not just explained?

Bring us the problem behind the article. We will tell you what it takes, what it costs, and whether we are the right team.

Test the page on first load, not after caching. Open it on a slow connection, disable browser memory, and watch what moves. Record the screen if you can; the eye catches shift faster than any report. If an element moves after a second, the user saw it move, and that is what CLS measures. The RTL mistakes that break Arabic interfaces explains the cases we see before launch.

INP and the scripts you did not notice

The worst INP problem is not in the code you wrote; it is in the tags you added. An analytics tool, a chat widget, a heatmap, an experimentation tool, each requests its own resources and competes with interaction. Do not delete them all; ask about each one: what decision will its data change? If there is no decision, it is cost without return.

Chat especially: if it loads a large library on first paint, it affects INP even when nobody uses it. Load it on click, or replace it with a button that opens a light window. Analytics can load after the first interaction, or server-side where that is permitted. Every small deferral shows in the user experience before it shows in the analytics dashboard.

The measure-and-fix checklist

Use this list in order. Do not start optimising before you know which page, which metric and which audience.

  1. Separate Arabic: build a separate report for the Arabic version, and do not mix it with English.

  2. Read the 75th percentile: do not look at the average, but at 75% of real visits.

  3. Pick the pages: start with pages that bring visits or conversions, not the whole site.

  4. Test a real phone: a mid-range device, a slow network, and a first load with no cache.

  5. Review the font: two weights, preload, and test the swap on Arabic text.

  6. Review images: explicit dimensions, a modern format, and priority for the hero image only.

  7. Review tags: every third-party script needs a decision and an owner, or it is cost.

  8. Measure again: two weeks after the change, not an hour, and record the before and after.

Frequently asked questions

Are Core Web Vitals a ranking factor? They are a user-experience signal, not the only factor. Content, intent and links still matter more. But speed affects conversion directly, and that is reason enough.

Is the Arabic font always the cause? No, but it is a frequent cause. Test the font alone: load the page with the system font and compare LCP and CLS.

Is Lighthouse enough? It is a starting point, but it measures lab conditions. Real user data at the 75th percentile is the judge.

Should we remove every script? No. Keep what changes a decision, defer what can be deferred, and remove what nobody owns.

How long does optimisation take? It depends on size. Start with the three largest pages, measure the effect, then move on. Continuous improvement beats a one-off project.

Do you help with measurement or with fixes? Both: we read the data, identify the cause, and fix the code, font, images and scripts.

Sources

Cover image: Tirza van Dijk tirzavandijk, CC0, via Wikimedia Commons.

Get in Touch

We'd love to hear from you. Fill out the form and we'll get back to you soon.