The short version
- Arabic search is one of the least competitive large markets on the web — the same effort buys a far better position than in English.
- Write the body in Modern Standard Arabic, but put the dialect phrasing your market types into the title, one heading and the FAQ.
- Drop diacritics in body copy. Almost nobody types them, and heavy vocalisation can fail to match the query entirely.
- lang="ar" without dir="rtl", and font stacks with no Arabic glyphs, are the two most common technical defects.
- Never redirect between language versions by IP. It breaks the canonical a crawler was sent to index and takes the hreflang cluster with it.
Arabic is among the most-spoken languages online and among the least well served by content. In most commercial categories the Arabic results are thinner, older and more poorly structured than the English ones — which means the same effort buys a much better position. The catch is that a handful of language and technical issues break Arabic sites in ways that never come up in English, and they are usually invisible to the team shipping the site.
1. Dialect versus Modern Standard Arabic
People write formally and search colloquially. A page written entirely in Modern Standard Arabic can be perfectly correct and still miss the phrasing its audience actually types, because search queries are frequently in Egyptian, Gulf or Levantine dialect — and dialects differ on ordinary commercial words, not just slang.
The workable compromise: write the body in MSA, which every Arabic reader understands, but make sure the dialect phrasing your market uses appears where it counts — in the title, in a heading, and in the question-and-answer section. Do not write an entire page in one dialect unless you sell in exactly one country.
2. Keyword research needs different sources
Arabic volume data in Western keyword tools is unreliable — often missing, often aggregated across spellings that behave differently. More useful sources:
- Autocomplete, typed in the dialect of the country you sell in, from that country.
- "People also ask" boxes, which are populated from real Arabic queries.
- Your own Search Console query list, which is ground truth and includes spellings you would never have guessed.
- Regional forums and social platforms — the phrasing there is what people type.
Expect to find several written forms of the same word. Where the variants are genuinely different words, they may need different pages; where they are spelling variants of one word, they belong on one page and the engine will usually reconcile them.
3. Orthography: hamza, alef, taa marbuta, diacritics
Arabic has several places where writers legitimately differ: أ / ا, ة / ه at word end, ى / ي. Search engines normalise much of this, but not all of it, and inconsistency within your own site is worse than picking either convention. Choose one convention, apply it everywhere, and keep it consistent between titles, headings and body.
Diacritics (tashkīl) should be omitted in body copy. Almost nobody types them, and a page written with full vocalisation may not match the undiacritised query at all. Use them only where a word is genuinely ambiguous.
4. The technical layer that RTL breaks
- html lang and dir must both be set — lang="ar" dir="rtl". Setting one without the other is the most common single defect on Arabic sites.
- The font must contain Arabic glyphs. A Latin-only font stack falls back per-character and produces text that renders inconsistently, or as boxes on some devices.
- Mixed-direction text — Latin brand names, numbers, URLs inside Arabic sentences — needs correct bidirectional handling, or punctuation lands on the wrong side of the line.
- URLs: transliterate rather than percent-encoding Arabic. Encoded Arabic slugs are technically valid but unreadable when shared, and unreadable links are shared less.
- Titles and meta descriptions: Arabic conveys more per character than English, so the character limits behave differently. Write to the visible snippet, not to a character count copied from an English guide.
- Layout: mirror the whole interface, not just the text. A right-aligned paragraph inside a left-to-right layout reads as broken.
5. hreflang, and the redirect that destroys it
If the site exists in Arabic and another language, each page needs a self-referencing canonical at its own language URL plus a complete bidirectional hreflang cluster — every URL listing every other, including itself, plus x-default. The Arabic page is not a duplicate of the English one; canonicalising it to English asks the engine to drop it, which is the opposite of the intent.
And the thing that quietly undoes all of it: redirecting visitors between language versions based on IP address or Accept-Language. A crawler bounced off the URL it was told to index treats that URL as unusable, and the cluster degrades. Detect language for first-time humans if you want, but never redirect a crawler, and never redirect away from a URL that is somebody's canonical.
6. Translated content underperforms — and it is visible
Machine-translating an English article into Arabic produces text that is grammatical and lifeless, uses terminology no practitioner in the market uses, and answers questions the Arabic-speaking audience was not asking. It also inherits the English article's structure, which was shaped by English search results.
Content written for the Arabic query from the start does better, and the gap is largest exactly where it matters — commercial pages, where the terminology, the payment methods, the delivery expectations and the objections all differ by market. If translation is the only option, translate the topic, not the text: take the subject, redo the query research in Arabic, and rewrite.
7. Local signals matter more than in English
- Name the country and city explicitly. Arabic search is heavily country-segmented and pan-Arab pages compete badly against local ones.
- Prices in local currency, on the page as text — not only inside a script-rendered widget.
- Local payment methods and delivery times stated plainly; these are the objections that decide the sale.
- A local phone format and address if you have one, and a Google Business Profile if you serve a physical area.
Why it is worth doing
In English, ranking for a commercial query means beating fifty companies that have been optimising for a decade. In Arabic, in many of the same categories, it means beating five pages that were written once and never updated. The technical work is a checklist — much of it the same checklist any store should run. The content work is genuinely writing for the market rather than translating at it. Both are ordinary — the advantage comes from how few competitors have bothered.
Frequently asked questions
Should I write in Modern Standard Arabic or dialect?
Body copy in MSA so every Arabic reader understands it, with the dialect phrasing your market actually searches in the title, a heading and the FAQ. Full-dialect pages only make sense when you sell in one country.
Do diacritics help Arabic SEO?
No — they usually hurt. Almost nobody types them, and heavily vocalised text can fail to match the undiacritised query. Use them only to disambiguate a genuinely ambiguous word.
Can I use Arabic words in URLs?
They work, but they percent-encode into an unreadable string when copied or shared. Transliterated Latin slugs are more practical and lose nothing in ranking.
Is translated content good enough for Arabic SEO?
Rarely. Translation inherits the English article's structure and terminology and answers questions the Arabic audience did not ask. Redo the query research in Arabic and write for those queries instead.
What is the most common technical mistake on Arabic sites?
Setting lang="ar" without dir="rtl" (or the reverse), and using a font stack with no Arabic glyphs. Close behind: IP-based language redirects, which break the canonical a crawler was sent to index.