Google has an explicit position on automatically translated content: it considers it spam when the sole objective is to manipulate rankings. But the line between “acceptable translation” and “penalisable content” has more nuance than is usually explained.
This guide covers how to implement an Arabic version of your website so that Google indexes it correctly and ranks it.
What Google actually says about automatic translation
Google’s Search Quality Rater Guidelines categorise as “Lowest Quality” pages that contain “automatically translated content without human review or curation”.
The criterion is not whether AI or automatic translation was used — it is whether the result is useful for the user. An automatically translated text reviewed and adapted by a native speaker can be perfectly valid. An AI-generated Arabic text without review, with grammatical errors, syntactic calques from English and vocabulary inappropriate for the market, is exactly what Google wants to filter out.
The real risk: duplicate content between versions
The most common risk is not penalisation for “automatic translation” but duplicate content between languages.
When a website launches an Arabic version with the same content as the English version translated, Google may:
- Not index the Arabic version, considering it a duplicate
- Index it but not rank it because it adds no differential value
- Cannibalise between versions if hreflang is incorrectly configured
The solution is not necessarily to write completely different content — it is to adapt the approach to the market. An Arabic user searching for “شراء زيت زيتون إسباني” (buy Spanish olive oil) has different questions to a European user.
Correct technical structure for a bilingual EN+AR website
Option 1: language directories (recommended)
yourdomain.com/en/ → EN version
yourdomain.com/ar/ → AR version
Advantages: everything on one domain, authority is shared, easier to maintain.
Option 2: subdomain
yourdomain.com/ → EN version
ar.yourdomain.com/ → AR version
Google treats subdomains as separate sites in terms of crawl, but authority signals do partially transfer.
Option 3: separate domain
Only makes sense when targeting a single market with its own ccTLD. There is no .ar for generic Arabic — ccTLDs are by country (.sa, .ae, .eg).
Implementing the lang and dir attributes
Each Arabic page needs:
<html lang="ar" dir="rtl">
<meta charset="UTF-8">
The dir="rtl" is necessary for the browser and Google to understand the reading direction. Without it, Arabic text may render left-to-right in some contexts.
Why AI translation without review damages rankings
There are concrete technical reasons beyond Google’s quality criterion:
1. Syntactic calques that confuse semantic analysis. Arabic has a VSO (Verb-Subject-Object) grammatical structure that differs from English SVO. A direct translation sounds strange in Arabic and affects how Google understands the main topic.
2. Keywords that are not used. The keyword the Arabic user searches for is not the literal translation of the English keyword. If the text uses terms an English speaker would use, the semantic density for relevant keywords is low.
3. Incorrect named entities. Google uses entities as signals of thematic relevance. Automatic translation may transliterate names in ways Google does not recognise.
4. Low engagement due to poor linguistic quality. An Arabic-speaking user who lands on a page with grammatically incorrect Arabic leaves quickly. Low time on page and high bounce rate are negative signals for the algorithm.
The correct implementation process
Step 1: native keyword research before translating. Before translating any page, you need to know what terms the Arabic user uses. The translated page must incorporate those terms naturally.
Step 2: AI-based translation (optional but efficient). Using GPT-4, DeepL or Google Translate as a first layer is acceptable if the next step is native review.
Step 3: review and adaptation by a native of the target market. The reviewer must correct grammatical errors, adapt vocabulary to the market register and verify that native keywords are incorporated naturally.
Step 4: technical configuration. lang="ar" and dir="rtl" in the HTML, bidirectional hreflang between EN and AR versions, correct canonical on each version, updated sitemap.
Step 5: indexing and monitoring. Request indexing in Google Search Console and monitor positions for Arabic keywords filtering by country SA, AE, EG.
Content that should not be translated without adaptation
Testimonials and case studies: Arabic users value social proof from clients in the same market. A testimonial from a European client does not carry the same weight.
CTAs and microcopy: the register and way of inviting action vary by market. In the Gulf, a more formal register is used. In Egypt, more direct and colloquial.
Data and statistics: if the text cites European market studies, translating them without adapting to the Arabic context generates distrust.
If you want to implement the Arabic version of your website correctly from the start, request your free audit.