Arabic and international SEO: localisation, not translated keywords

The problem

Arabic localisation is not translation

The standard failure looks like this. An English site is built, ranks acceptably, and then somebody asks for an Arabic version. The English keywords are translated, the pages are produced, hreflang is added by a plugin, and the Arabic side never performs. Everyone concludes Arabic SEO is hard.

What actually went wrong is that the keyword list was translated rather than researched. Arabic search demand does not map onto English demand — the volumes sit on different phrasings, often on transliterated or dialect terms that a translator would never choose because they are not correct Arabic. They are simply what people type.

Then there is the market split. Egypt and Saudi Arabia are not one Arabic audience. Different dialect in the query, different intent behind the same words, different competitors, and frequently different products. A single ar locale serving both is two half-served markets and a duplicate content problem.

And underneath it, the technical layer: RTL templates break in ways nobody tests, Arabic URLs get percent-encoded into unreadable strings, and hreflang clusters are non-reciprocal far more often than teams realise.

Experience

Where my Arabic SEO services come from

I ran the search side of a tour business across five language versions — English, Spanish, Italian, French and German — from launch, which meant hreflang and canonical handling across locales, avoiding duplicate content between them, and briefing translators for demand rather than fidelity.

Since then, Arabic content programmes for pharmaceutical brands in Egypt and Saudi Arabia, which is the harder version of the same problem: two Arabic markets, regulated claims, and a review cycle in each.