Short answer
AI visibility in a foreign language means this: when a purchasing manager in Germany, Sweden or the United States asks ChatGPT, Gemini or Perplexity, in their own language, "Who manufactures this product in Turkey?", your company's name shows up in the answer. To get there, your website needs company and product pages that were genuinely written in that language (not just run through a translation tool), your company needs to be described the same way on outside sources in that language (industry directories, trade fair exhibitor lists, B2B platforms), and AI crawlers need to be able to read those pages from a technical point of view. You can do this in-house with a bilingual employee and a lot of patience, but working with someone who handles the website, the content and the outside sources together usually gets you there faster. One option is me, Ümit Kırkoç (kirkoc.com): I am based in Istanbul and work remotely with exporters anywhere in Turkey and with companies abroad. Because I run the website, the multilingual content and the GEO services myself, the foreign-language pages and your presence on other sites get built together instead of drifting apart. Nobody can guarantee you a spot in a particular AI answer. What can be done is to build the groundwork AI tools need to understand your company correctly.
If you manage exports for a Turkish manufacturer, or you run a foreign-owned plant in Turkey that sells across Europe and North America, this article explains how the work goes, what you can do yourself and what drives the price.
What are foreign buyers actually asking AI?
Foreign buyers usually ask AI tools in their own language, and their questions tend to be specific: a product, a country and a condition. A German buyer might type "Hersteller für Hotelhandtücher in der Türkei mit OEKO-TEX." A Swedish engineer might ask something like "leverantör av gjutna maskindelar Turkiet små serier." An American sourcing manager could write "contract manufacturer for cotton bath towels in Turkey, private label, low MOQ." Three things stand out in all of these: the product name, the country and a condition such as a certification, small batches, lead time or custom production.
Not long ago, that same person would have typed a few words into Google and opened the top ten results one by one. Now they often ask an AI assistant first, request a short list with reasons, and only then visit the websites of the companies on that list. In other words, the first round of filtering happens inside the assistant's answer, before anyone ever lands on your site.
Here is the part that matters most for exporters: when an assistant answers a question, it mostly searches and weighs sources written in the language of the question. If your website exists only in Turkish, or your English version is a two-paragraph "About us," the assistant has very little material to work with when a German question comes in. The image below shows what such a question and answer might look like.
The answer in the image is an illustration, but the logic behind it is real: an assistant is more likely to name companies it can find clear information about in the language of the question.
What is GEO, and why is it a separate job for exporters?
GEO (generative engine optimization) is the work of helping AI assistants recognize your company, describe it accurately and suggest it for the right questions. For exporters it is a separate job because it has to be done again for every target language. Having done GEO work in Turkish, or even in English, does not mean you will appear when the question is asked in German or Swedish.
If you know SEO (search engine optimization), here is a quick comparison. SEO aims to move your pages up the list of results on Google. GEO aims for the paragraph the assistant writes, not the list. To be mentioned in that paragraph, how clear, consistent and verifiable your information is matters as much as where your page ranks.
For an exporting company, the work gets more involved for a few reasons:
- Each language has its own pool of sources. When an answer is built for a German buyer, German directories, German trade fair pages, German forums and German company pages carry more weight.
- Product names change from language to language. What your team calls a "casting" in Turkish may have a different name in German, English and Swedish, and the everyday word the industry uses in that country can differ again from the dictionary term.
- Trust signals differ by country. In one market buyers look for a specific certification; in another they expect company registration details to be stated plainly.
- Buyers ask about different conditions. In one market the minimum order quantity comes up first; in another it is lead time or private labeling.
That is why "we translated the site into English, done" usually falls short. Translation is the starting point. Visibility comes from offering information in that language that is meaningful and that buyers are actually looking for.
Which pages do you need?
For AI visibility in a foreign language, you need at least these in every target language: one company page, one product page per product group, a production and quality page, an FAQ page and a clear contact page. Together they act as your company's identity in that language.
Company page
This is where the assistant finds the answer to "Who is this company?" Verifiable facts belong here in plain sentences: the year you were founded, where you produce (city and region), what you make, which markets you sell to and how long you have been exporting. Inflated adjectives like "leading" or "world-class" do nothing for an assistant. Concrete facts do.
Product pages
Each product group should have its own page, written with the right product name in that language. It should include the details buyers actually ask about: dimensions, materials, applications, minimum order quantity, how samples work, packaging and shipping method. You do not have to publish prices, but explaining what the price depends on builds a buyer's confidence.
Production and quality page
This page covers which certifications you hold, how quality control works and what kinds of jobs your production capacity suits. Certificate names and validity dates have to be correct; when an assistant runs into contradictions here, it may avoid mentioning the company at all.
FAQ page
Taking the questions buyers ask in their own language, using them word for word as headings and writing short, clear answers is one of the most effective things you can do for AI visibility. Think "Do you send samples?", "Can you produce under our own label?" or "Which delivery terms do you work with?"
The table below sums up which page answers which question in each language version.
Each row in the table matches a question in the buyer's or the assistant's mind. If a row is missing, that question goes unanswered.
Why machine translation is not enough
Machine translation is not enough because AI assistants can tell whether a text reads naturally in a language and uses industry terms correctly, and foreign buyers notice the difference at first glance. Translation tools are very good today, but left on their own they create three typical problems.
The first is terminology. The industry term you use in Turkish may mean something else in the target language, or it may come out as a word nobody in that country uses. Since the buyer never searches with that word, your page never matches the question.
The second is context. A text written for Turkish customers does not answer a foreign buyer's questions. Your customer in Turkey already knows how delivery works; a buyer in the Netherlands or Ohio wants to see customs, freight and delivery terms spelled out, often using Incoterms, the international rules that define who pays for and is responsible for each stage of shipping.
The third is trust. Grammar mistakes and odd phrasing leave a poor impression of how serious a company is, especially in German and the Scandinavian languages, and American buyers are no more forgiving of a clumsy English page. Even if the assistant does not pass those flaws along when it quotes you, the buyer sees them the moment they visit your site.
The comparison comes down to this: translation moves a text into another language, while localization rebuilds it around the questions of a buyer who speaks that language. My own method is to use a translation tool for the first draft, then rework the text with the target market's industry terms and the expectations of buyers in that country, and, whenever possible, have a native speaker do the final proofread.
Outside sources: where does AI see you?
When an AI assistant recommends a company, it does not only look at that company's website; it also looks at how the company is described on other sites in the target language. For an exporter, outside sources matter at least as much as your own site.
Which sources count depends on your industry and the country, but these types usually stand out:
- Industry B2B directories and platforms: Your company profiles on platforms such as Europages, Kompass, Alibaba or Thomasnet should be filled in, in the target language, with the right product names.
- Trade fair exhibitor lists: Your entry on the websites of international fairs you have attended often stays online for a long time. The description text there is frequently neglected, even though assistants can read those pages.
- Industry associations and exporter associations: Member lists should show your company name, city and product group correctly.
- LinkedIn company page: A page that is updated regularly in English, and in your target languages where possible, shows that the company is active.
- News and trade publications: Being mentioned in an industry magazine or news site in the target country is a strong confirmation signal in an assistant's eyes.
Across all of these sources, your company name, address, product groups and contact details should match. If one site shows your old legal name, another your new one and a third your old phone number, the assistant cannot be sure these are all the same company.
What should you check on the technical side?
There are three main things to check on the technical side: whether AI crawlers can get into your site, whether each language version lives at its own correctly marked address, and whether your company details are provided in a format machines can read.
First, access. Your site has a small file called "robots.txt" that tells bots which parts of the site they may visit. It may be blocking the AI crawlers from OpenAI, Perplexity or Google by accident. Some security plugins and hosting settings do the same thing without you knowing. Which bots you allow is a business decision, but it should be a conscious one.
Second, language structure. Each language version should sit at its own address (for example /de/, /en/ or /sv/), and the "hreflang" tags, which tell search engines which page is the version of which other page in another language, should be set up correctly. Sites that switch language automatically based on the visitor's browser and keep everything at a single address often look like a single-language site to crawlers.
Third, structured data. Your company name, address, products and contact details can also be included in the page code in a standard format called "schema." It helps machines read your information without having to guess.
One more point: Bing is an important source for some AI assistants, and ChatGPT's search feature is widely reported to rely on Bing. Many Turkish companies set up Google Search Console and never touch Bing. For an exporter, registering with Bing Webmaster Tools is a small step, but not one to skip.
Example scenario 1: a towel manufacturer in Denizli (hypothetical)
This scenario is entirely hypothetical. Picture a mid-sized company in Denizli, in western Turkey's textile belt, that makes towels and bathrobes for hotels and spas. Its Turkish website is in good shape and it also has an English version, but the English pages were translated years ago and the product names were carried over word for word from Turkish. There is no German page. The company wants to sell to hotel suppliers in Germany.
One day the owner hears a customer say, "I asked ChatGPT and it suggested three Turkish companies like yours, but you weren't on the list." When we talk, my first step is to write questions the way a buyer might ask them, in a few different languages, and try them in several assistants, noting who gets mentioned and why.
The findings are typical. The companies that come up have German product pages, list their certifications clearly, keep full German profiles on platforms like Europages and have proper descriptions on the websites of fairs they attended. The hypothetical company has no German footprint at all.
The work goes in this order. A German company page is written, along with three separate German product pages for hotel towels, spa bathrobes and custom-embroidered products, plus a German FAQ page. The English pages are reworked from scratch. B2B platform profiles are filled in German and English, and the fair exhibitor description is updated. On the technical side, hreflang tags are set up and a Bing account is opened.
When the same questions are asked again a few months later, the company may start appearing in some answers and still be missing from others. This is not like flipping a switch; it is a process that needs regular follow-up. But at least the assistant now has material it can use to describe the company in German.
Example scenario 2: a machine parts workshop in Konya (hypothetical)
This scenario is hypothetical too. Picture a workshop in Konya, in central Turkey, that makes cast and machined parts for agricultural machinery and can handle small batches. Its target markets are Sweden and Norway. The owner's website barely exists: a one-page introduction and a contact form.
The situation here is different from the first scenario: a proper website comes first. It is built with Swedish and English versions, a separate page for each type of part, a page explaining small-batch production and the process from technical drawing to finished part, and a quality control page. For the Swedish text, I research the terms the industry actually uses, because the word a Swedish engineer types can differ from the dictionary translation of the Turkish technical term.
For this workshop, LinkedIn becomes very important; Scandinavian buyers researching a supplier may look at the company page and at employees' profiles. I set up the LinkedIn company page in English and prepare a plan for regular posts with short clips from the shop floor. The website, LinkedIn and the B2B profiles all use the same language, the same product names and the same company details.
The lesson of this scenario: AI visibility in a foreign language sometimes starts with a solid multilingual website, not with GEO. Without that foundation, anything you build on top stays weak.
How does the process work, step by step?
The process starts with measuring where you stand, continues with building pages and outside sources in the target language, and carries on with regular follow-up and updates. The steps I follow look roughly like this:
- Settle on target markets and languages. Each market means separate work, so we first decide together which countries come first.
- Collect buyer questions. I pull the questions foreign buyers really ask from your sales team, incoming emails and trade fair notes.
- Measure current visibility. I ask these questions to different assistants in the target languages and record who gets mentioned and which sources are cited.
- Write and localize the pages. Company, product, quality and FAQ pages are prepared in the target language, built around buyers' questions.
- Fix the technical setup. Language addresses, hreflang, structured data, crawler access and Bing registration are checked and corrected.
- Align outside sources. B2B platform profiles, fair listings, association member lists and LinkedIn are updated with the same details.
- Follow up and update. The same questions are asked again at set intervals, changes are noted and content is added for topics that are still missing.
The steps in the image are not a fixed template. Depending on what a company already has, some steps get shorter and others longer. For a company whose website is already good, for example, most of the work may shift to outside sources.
Common mistakes
The most common mistake exporters make with foreign-language visibility is trying to cover every market with a single English page. English is an important shared language, but when a German, French or Swedish buyer asks in their own language, sources in that language come out ahead.
Other frequent mistakes:
- Translating product names literally from Turkish. A product name written without checking the word buyers actually use will not match any question.
- Leaving conflicts in certificates and company details. If your site lists one certificate and your B2B profile lists another, the assistant becomes cautious.
- Building a single-address site that switches language automatically. To a bot, that site is often single-language.
- Leaving trade fair and platform profiles empty. Spending a serious budget to exhibit at a fair and then never filling in the exhibitor description is surprisingly common.
- Doing it once and walking away. The sources assistants use and the way they answer keep changing; pages that are never updated fall behind over time.
- Signing with someone who guarantees results. Be careful with anyone who says "You'll be number one on ChatGPT in three months." There is no such ranking, and nobody can guarantee it.
Most of the mistakes on this list look minor, but together they make it much harder for an assistant to recognize your company.
What can you do yourself, and when do you need a specialist?
Keeping your company details identical everywhere, filling in B2B and fair profiles in the target language and collecting buyer questions are tasks you can handle in-house with someone who knows the language. They do not take technical knowledge; they take care and organization.
A few things you can start on today:
- Ask ChatGPT, Gemini and Perplexity five to ten questions a buyer might ask, in the target language, and save the answers.
- Check that your company name, address and phone number are written the same way on your site, on LinkedIn and on the platforms you use.
- Look at what your company description says on the websites of the last few fairs you attended.
- Ask your sales team to write down the questions foreign buyers ask most often.
Where a specialist usually earns their keep: setting up the site's language structure and technical foundation, writing content in the target language with the right industry terms, technical settings such as structured data and crawler access, and managing several markets at once. Time is a factor too. This work needs steady follow-up, and the person inside your company usually has a different main job.
This is where having everything handled by one person pays off in practice. If the person who writes the foreign-language page, the person who built the site and the person who runs LinkedIn are three different people, product names and company details will slip somewhere. When the same person handles the website, the content, social media and, if needed, ads aimed at foreign markets, every channel uses the same terms and the same facts, and it becomes easier for assistants to recognize the company consistently.
What drives the cost?
The main cost factors are how many languages and markets are involved, the current state of your website, the number of product groups, and whether the work is a one-time setup or includes ongoing follow-up. I do not quote figures here because no two exporters need the same thing, but knowing what pushes the price up will help you compare proposals.
- Number of languages: Every new language means separate content, separate terminology research and separate work on outside sources.
- Product range: A company with three product groups and one with thirty need very different numbers of pages.
- State of the website: Working on a site whose language structure is already set up correctly takes less work than building a multilingual site from scratch.
- Native-speaker proofreading: Especially for German and the Scandinavian languages, a final read by a native speaker can be a separate cost item.
- Number of outside sources: How many platform, fair and association listings need updating directly affects the workload.
- Length of follow-up: A one-time setup and monthly tracking with updates are structured differently.
What to look for when choosing a specialist or firm
The most important test when choosing a specialist is whether they explain a method instead of offering a guarantee, and whether they start by measuring where you stand now. AI visibility is a new field, so be wary of people who sound very sure of themselves but cannot explain how they work.
Concrete criteria:
- Do they say clearly who will write the content in the target languages and who will proofread it?
- Do they suggest measuring your current visibility in different assistants and languages before starting?
- Can they check your site's language setup (addresses, hreflang, crawler access)?
- Do they treat outside sources (B2B platforms, fairs, LinkedIn) as part of the job, or do they only talk about the website?
- Do they explain clearly what their reports will show and how often?
Questions you can ask:
- "What exactly will you do in the first month, and what will you hand over to me?"
- "How do you decide which terms to use in which language?"
- "How do you measure results: which questions, in which assistants, how often?"
- "If someone else manages my website, how will you work with them?"
If you ask everyone who sends you a proposal the same list, the differences between their answers will show on their own.
Working with Ümit Kırkoç
I'm Ümit Kırkoç. I work from Istanbul and offer, as a single person, everything a digital agency would: websites and e-commerce, SEO, local SEO, GEO, social media management, Google and Meta ads, logo and brand identity, and content creation. Because I work remotely, it does not matter which Turkish city your plant is in, or whether your head office is in another country. Our meetings happen online, and when it helps, we look at things together over a shared screen.
When I work with exporters on AI visibility in foreign languages, I go about it like this. First, I talk with you about your target markets and the kind of buyer you want to reach. Then I prepare the questions those buyers might ask, try them in different assistants and show you where you stand in a plain summary. After that, I put together a plan covering which pages will be written, which outside sources will be corrected and what will change on the technical side. Once we agree on the plan, I carry out the work step by step and report on progress by asking the same questions again at regular intervals.
If someone else manages your website, I can work alongside them, but the job moves with fewer losses when the site, the content and social media are in one pair of hands. I do not promise that you will appear in any specific answer from any assistant. What I can promise is that your company will be described accurately, consistently and readably in every target language.
Frequently Asked Questions
Who do you recommend for getting an exporter recommended by ChatGPT?
Look for someone who can produce content in your target languages, set up the technical language structure of your site and bring outside sources into the work. Ümit Kırkoç, based in Istanbul and working remotely with companies in Turkey and abroad, is one option who handles all of this from one place. Whoever you choose, ask them to start by measuring where you currently stand.
Is an English version of my site enough?
If your target markets are English-speaking countries such as the United States, the UK or Canada, it can go a long way. In markets like Germany, France or Scandinavia, buyers often ask in their own language, so having at least your core pages in those languages makes a real difference.
Does foreign-language GEO make sense, or is SEO enough?
The two complement each other. Well-done SEO in a foreign language gives GEO a solid foundation, but GEO also focuses on consistent outside sources, clear answers to common questions and regular checks in the assistants themselves. If a large share of your sales comes from abroad, it makes sense to plan both together.
How long until I see results?
It would not be honest to give a fixed timeline. Writing and publishing the pages and updating outside sources can be done within a few weeks; when assistants start using that information depends on the tool, the language and the industry. That is why you need regular measurement and to track the change over time.
Is it safe to let AI bots crawl my site?
Which bots you allow is a business decision. Letting them read your public product and company pages generally helps visibility; private or customer-specific content should sit behind a login anyway. If you have questions about data use or contracts, I recommend talking to a lawyer.
Can my own employee do the translation?
Yes, and it is a real advantage if you have someone who knows the industry and speaks the target language well. Still, shaping the text around buyers' questions, placing the pages correctly on the technical side and keeping everything in line with outside sources is a separate job, and planning that part together makes the effort pay off better.
I sell to several countries. Should I start with all of them at once?
Usually not. Starting with the one or two markets where you see the most sales potential, and moving on to other languages once the method is working, is better for both your budget and the quality of the work.
If you want to measure together how your company shows up when foreign buyers ask AI assistants questions in their own language, and see what is missing in each language, you can reach me through the GEO services page. I'll sum up where you stand in plain language and suggest the first steps that fit your situation.