Multilingual SEO and AI-search localization, with a native linguist on every job
People search in their own words, and those words are often not a translation of yours. We research how people in your target markets search for what you offer, in the languages and Arabic dialects we list. AI drafts your localized pages and metadata around the keywords you approve, and native linguists edit them at the service level you choose. If you want, we also report what AI assistants say about your brand in each language, before your localized pages go live, or on a date we agree with you, and again later.
Keyword sheet · English → ArabicExample
Source keywordphone caseIntent: buying
-
Saudi ArabiaSaudi usagehreflang ar-SA
كفرات جوال
Everyday term, in the plural for the Saudi store page. The keyword keeps جوال because كفرات alone can mean car tires.
-
EgyptEgyptian usagehreflang ar-EG
جراب موبايل
Everyday term. Use on the Egyptian store page.
-
Several marketsModern Standard Arabichreflang ar
غطاء هاتف
Formal term, understood across markets. Use on a single Arabic page for several markets.
- Title tag, Saudi page
- Linguist
كفرات جوال بتصاميم متنوعة | تيلورا
In English: Phone cases in a range of designs | Tilora
- Issue log
- Saudi title · Terminology · Minor · The AI draft used غطاء هاتف, the formal term. The linguist changed it to كفرات جوال, the Saudi term in your approved keyword sheet.
What you get
- Keyword research for each market. Native linguists research how people in each target market search for what you offer: the words, spellings, and questions they use, and the intent behind each search. For Arabic, that includes the difference between Modern Standard Arabic and the words people type in each dialect, Arabic written in Latin letters (Arabizi) where people in a market search that way, and searches that mix Arabic with English brand and product names.
- Localized pages and metadata. Titles, meta descriptions, headings, URL slugs, image alt text, and page copy, localized around the approved keywords and edited by native linguists at your service level. If a page is already live, we flag any change to its URL slug, so your developers can redirect the old address. Text in your structured data, such as product names and descriptions, and breadcrumb names, is localized to match the text on the page.
- An hreflang plan. Our recommendations for the hreflang annotations that connect your language versions: the language and region code for each page, the links between versions, a canonical URL in the same language for each page, a default page for other visitors if you want one, and notes on letting visitors switch languages without redirecting them automatically.
- An AI answer snapshot. We ask the AI assistants you choose a set of questions you approve, in each language, and record what they say about your brand, before your localized pages go live, or on a date we agree with you, and again after an interval we agree with you.
- An issue log. What our checks and linguists found and fixed in each file.
What this service doesn’t cover
Search engines and AI companies decide where a page appears in search results and what an AI assistant says, and both change over time. We don’t promise either. We deliver research, localized content, and reports of what we found.
We don’t build links, buy placements, or publish content on other websites. Technical audits of site speed, crawling, and indexing are outside this service. We don’t change your website: your team or your developers publish the localized files and put the hreflang plan in place.
How it works
Every step has a named owner, and a native linguist works on every job.
- Intake. You send the pages to localize, as links or an export from your content management system (CMS), and tell us the target markets, your priority products or services, and any keywords you already target. We agree on the markets and, for Arabic, the variant for each country. If your order includes the AI answer snapshot, we agree on the questions and the assistants, and run the first round before anything changes.
- Keyword research. For each market, native linguists of the target language or dialect list the ways people there name and ask about what you offer. Candidates come from your source keywords, your published content, search data you share with us, and public search evidence for that market, such as the searches a search engine suggests and the pages that already appear for a term. We check this evidence for that market’s country and language, and the keyword sheet records how. Agents help gather candidates from your content. Each keyword gets its source, its intent (looking for information, comparing options, or buying), and a note on dialect and spelling. A native linguist approves the keyword sheet, and then you approve it. Approved keywords join your glossary, marked with their market and the pages they apply to.
- AI translation. AI translates your pages and metadata against your glossary, the approved keywords, and your style rules.
- Linguist editing. A native linguist edits every title, description, heading, URL slug, alt text, and segment of page copy, and places the keywords where they read naturally. For Arabic URL slugs, we agree with you whether to use Arabic words, which are percent-encoded in links, or Latin letters. A keyword that reads badly in a sentence is not forced into it: the linguist records the choice in the issue log.
- Checks and review. QA agents check accuracy, terminology, and style. At AI + Edit, a second linguist checks a sample; at AI + Edit + Review, a second linguist reviews every segment.
- Quality gate and delivery. Automatic checks run on every file: the length limits in your spec for titles and descriptions, or limits we agree with you if you have none; the approved keyword in each element you agreed on; URL slug format; valid language and region codes in the hreflang plan, with each page listing itself and every other version; and tags and placeholders. The linguist settles every flag by fixing it or recording why it stays, and if a file fails the gate, it does not ship.
This is our AI cycle applied to search: agents do the volume work of gathering candidates and translating, and a native linguist decides every keyword and every line.
The AI answer snapshot
The snapshot records what AI assistants say about your brand in each language. It is descriptive: it reports what we found, and it makes no prediction.
- The questions. We agree on a set of questions with you for each language: what buyers ask about your category, your brand, and your products, phrased the way people in each market ask them. The questions use only public information, such as your brand and product names. Your files and unpublished content never go into them, and you approve the final questions before we ask them.
- The assistants. The AI assistants you choose, used only through business AI services, such as a business plan or API access, whose terms allow business use and bar training on what we send. We confirm before we quote which of them we can use this way in the languages you need, and we don’t include one we can’t. We ask each question in a new conversation, with saved memory and personalization turned off where the assistant allows it.
- Repeated runs. Answers can change from one run to the next, so we ask each question several times in each round and report how often the answers mention your brand and how they describe you. Your quote states how many runs we make.
- What we record. For each answer: the date and time; the assistant and, where it shows one, its version; the language, location, and account settings we used; the full text of the answer; and any sources or links it shows. A native linguist notes whether the answer mentions your brand and how it describes you, and flags statements that contradict your published content, terms that don’t match your glossary, and answers in the wrong language or dialect.
- Before and after. We run the same questions before your localized pages go live, or on a date we agree with you, and again after an interval we agree with you. A difference between rounds does not, by itself, show what caused it: answers vary between runs, and assistants change their models and sources over time. The report describes what changed in the answers, not why.
- What it can’t show. Assistants answer differently by account, location, time, and version. Business and API access can also answer differently from the free versions many people use, and we ask from our own location unless an assistant lets us choose another one. The snapshot shows what the assistants said in our sessions on the dates in the report: a sample, not a measure of what every user sees.
How search terms differ by market
- Arabic: one product, three keywords. A phone case is usually كفر جوال in Saudi Arabia and جراب موبايل in Egypt. A common Modern Standard Arabic term is غطاء هاتف. A page that uses only the Modern Standard Arabic term can miss the words shoppers in each market type.
- Words with a second meaning. In Saudi Arabia, كفر on its own also means a car tire, and كفرات car tires, so the keyword keeps جوال. Linguists check what else a candidate keyword means in each market before it goes on the sheet.
- Spelling variants. People type the same Arabic word in more than one way: with or without hamza, as in أغطية and اغطية, with ة or ه at the end of a word, and with ى or ي. The keyword sheet lists the forms we find and the one your page uses.
- Arabizi and mixed searches. Some people type Arabic in Latin letters, or mix Arabic with English brand and model names. Where people in a market search that way, the keyword sheet lists those forms and notes which ones belong in your page text and which ones belong only in your research.
- Spanish for Spain and Latin America. A mobile phone is usually a “móvil” in Spain and a “celular” in Latin America, so the keywords for a phone case differ too. Latin American Spanish covers many countries: tell us which ones matter most.
- Codes name a language and a country, not a dialect. An hreflang code names a language and, if you add one, a country: an Arabic page for Saudi Arabia is ar-SA, and an Arabic page for Egypt is ar-EG. Latin American Spanish has no single country code, so the plan maps it to the countries you target, such as es-MX for Mexico, and one page can carry several of those codes.
How deep the linguist’s work goes
SEO and AI-search localization runs on two of our three service levels. Your quote names the level for each type of page.
| Service level | Keywords | Pages and metadata |
|---|---|---|
| AI + Edit | Every keyword approved by a native linguist | Every title, description, heading, URL slug, alt text, and segment of page copy edited by a native linguist, then a sample checked by a second linguist |
| AI + Edit + Review | Every keyword approved by a native linguist, then reviewed by a second linguist | Every segment edited, then reviewed by a second linguist |
We don’t run this service at AI + Check. Titles, descriptions, and headings are the headlines of your pages, and AI + Check never runs on headlines. AI + Edit is the usual level for product, category, and support pages. AI + Edit + Review fits campaign and brand pages. Pages with regulated or high-risk content, such as health, legal, or financial information, run at AI + Edit + Review or without AI.
Languages and dialects
We research keywords and localize pages in the languages we list, including Modern Standard Arabic and, on request, Egyptian, Gulf, Levantine, and Maghrebi Arabic. Other languages on request: we confirm we can run them before we quote.
Search terms can differ between countries that share a language or a dialect, as they do across the Gulf. Tell us which countries you target in each language, and your quote names the variant for each one.
What you send and what you get back
You send
- The pages to localize, as links or a CMS export, or the files your pages are built from
- Your target markets, and for Arabic, the countries you target
- Your priority products, services, or topics
- Keywords you already target, and search data from your own tools, such as a report of the searches that already bring visitors to your pages, by country, if you have them, with personal data removed
- Your glossary and style guide, if you have them
- For the AI answer snapshot: your brand and product names, the questions that matter to you, and the assistants to include
You get back
- A keyword sheet for each market: each keyword with its source, its intent, notes on dialect and spelling, the page it belongs to, and the priority we agree with you
- Localized pages and metadata in your format: titles, meta descriptions, headings, URL slugs, alt text, page copy, and structured-data text
- An hreflang plan for your language versions
- The quality-gate result for each file
- An AI answer snapshot report for each language and assistant, before and after, if your order includes it
- An issue log, plus the quality reports your service level includes
- Glossary changes, including the approved keywords, and a query log
- Your translation memory (TM), updated, with a Translation Memory eXchange (TMX) file on request
How we quote SEO and AI-search localization
We quote per page for localized pages and metadata, and per keyword set for keyword research: one set for each page or topic in each market. The quote depends on the number of pages and markets, how much copy each page has, the Arabic variants, whether you need keyword research only or localized pages too, how many questions, runs, and assistants the AI answer snapshot covers, and the service level. Keyword research and the AI answer snapshot can each be ordered on their own. The quote states the basis, and we confirm any change with you before work continues.
Your content and data
- We sign your NDA, and your data processing agreement (DPA) if you have one, before you share files. If a clause doesn’t match how we work, we agree on the change with you before work starts.
- Pages that aren’t published yet can come as files. Please don’t put passwords in our forms: if a link needs access, we arrange it by email.
- When we use AI on your content, we use only business AI services whose terms bar training on it. We don’t use your content to train or tune our own AI tools.
- For the AI answer snapshot, we send only the questions you approve, which use public information, to the assistants you choose, through business AI services whose terms bar training on what we send.
- Prefer no AI? Native linguists then research your keywords and write your pages and metadata with no AI on our side, and the quote says so. The AI answer snapshot asks AI assistants about your brand, so we run it only if you ask for it.
- We delete your files on request, unless the law requires us to keep them. Files you upload through our quote form are deleted automatically from our form storage 90 days after upload.
Start with a free keyword sample
Send us the link to one page of your organization’s website, or of a website you’re authorized to work on, or the page as a file, and tell us one target market. A native linguist researches up to 10 keywords for that page in one target language, or one Arabic variant, and edits a localized title and meta description around them, drafted by AI, at AI + Edit. If you choose No AI on the form, or the page has regulated or high-risk content, such as health, legal, or financial information, the linguist does the sample with no AI on our side. You get the keyword sheet, with the source, intent, and dialect notes for each keyword, and the localized title and meta description. The sample covers only that page and market, and its findings are indicative. It doesn’t include the AI answer snapshot. One free keyword sample per company.
On the quote form, choose “SEO and AI-search localization” and the free keyword sample, and add the page as a link or a file. Under Target market or locale, tell us the market, and for Arabic, the country. We confirm the details by email before we start.
Questions about SEO and AI-search localization
Do you promise better search results or mentions in AI answers?
No. Search engines and AI companies decide where a page appears and what an assistant says, and both change over time. We deliver keyword research, localized pages and metadata, an hreflang plan, and reports of what we found, so your team can see what was done and judge the results with its own data.
Isn’t this the same as translating our keywords?
No. A translated keyword is often not what people in the target market type. We start from how people in each market search, and a native linguist decides each keyword. Your source keyword is a starting point, not the answer.
Which Arabic should our pages use?
We localize Arabic pages into Modern Standard Arabic by default. For a page written for one market, such as a store page for Saudi Arabia or Egypt, we can localize it into Egyptian, Gulf, Levantine, or Maghrebi Arabic on request. Many store pages keep Modern Standard Arabic copy and use the local term in titles and keywords, as in the example at the top of this page. Either way, the keyword sheet shows the terms people use in each market, and you choose the variant for each page.
Do you report search volumes?
The keyword sheet gives the source and intent of every keyword. If you have search-volume data from your own tools, send it, and the sheet shows it with its source. We don’t add volume estimates of our own.
Which AI assistants do you check?
The ones you choose, among those we can use through business AI services, such as a business plan or API access, whose terms allow business use and bar training on what we send. We confirm before we quote which ones we can use in the languages you need. The report names each assistant, how we reached it, and the date and time of every answer.
Can you publish the changes on our site?
We deliver localized pages and metadata in your format, such as a CMS export, XLIFF, or a spreadsheet, or we work inside your translation management system (TMS). Your team or your developers publish the pages and put the hreflang plan in place.