How to Make Your Multilingual Website Visible in ChatGPT Search and Perplexity
AI search doesn't crawl, rank, and reward the way Google does. If your multilingual site is invisible inside ChatGPT and Perplexity right now, it's not because you're doing SEO wrong — it's because you're still doing SEO.
No card details
Six months ago, a SaaS founder I was advising sent me a screenshot. He’d asked ChatGPT to recommend translation tools for a Shopify store. His company — a real, ranking, traffic-earning brand on Google — wasn’t in the answer. Two competitors were. One of them was a year old.
That’s the moment most founders realize the rules have changed. The traffic still flows on Google, but the recommendation layer is moving somewhere else. ChatGPT, Perplexity, Gemini, Claude — these systems decide which brands get mentioned when someone asks a buying question. And they don’t decide it the way Google does.
If you run a multilingual website, the problem compounds. You’re optimizing for ten search engines at once, in a dozen languages, and most of the visibility advice you’ve read was written for a world where blue links still mattered. Here’s what actually works in 2026.
Why traditional SEO doesn't translate to AI search
Google indexes pages. ChatGPT and Perplexity build answers. That distinction sounds small until you watch how it changes the rules of visibility.
When a user asks Perplexity “what’s the best website translation software for Shopify,” the system runs a live search, pulls in the top results, reads them, and synthesizes an answer. It cites three to five sources in that answer. Those citations are the new traffic. If you’re not one of them, you don’t exist for that query — even if you’re position #4 on Google.
ChatGPT’s behavior is similar but with a twist. Its responses pull from training data plus live browsing (when enabled). That means brands frequently mentioned across the open web — in articles, comparisons, Reddit threads, podcast transcripts — get surfaced more often, regardless of their classical SEO performance.
Three things shift when you understand this:
- Quantity of pages matters less. Quality of mentions matters more. A brand referenced in 40 long-form comparison articles will outperform a brand with 400 thin pages every time.
- Structured, scannable content wins. AI systems extract claims, statistics, and definitions far more reliably than they extract narrative prose. Articles that read like reference material get cited; articles that read like blog posts get ignored.
- Language coverage compounds. AI Overviews, ChatGPT, and Perplexity all respond in the user’s language. If you only exist in English, you’re competing for a fraction of the queries you could be winning.
That last point is the one most multilingual websites get wrong. They translate their site for users, then assume AI search will figure out the rest. It won’t. AI search needs explicit, well-structured signals that your translated content is canonical, intentional, and authoritative.
What ChatGPT and Perplexity actually look for
Anthropic, OpenAI, and Perplexity have all published varying degrees of detail about how their retrieval works. After running our own tests across 200+ queries in five languages, here’s the pattern that emerged.
1. Clear topical authority in each language version
AI systems weight sources by how concentrated their expertise is. A site that publishes one article on website translation in Japanese carries less authority than a site with twenty articles on Japanese localization, all interlinked, all consistent in voice.
Translating five blog posts into ten languages is worse, not better, than translating fifty blog posts into three languages. Depth signals expertise. Spread signals nothing.
2. Citable, fact-dense formatting
When ChatGPT or Perplexity reads your page, it’s looking for extractable claims. Sentences like “hreflang tags must be reciprocal” or “Japanese e-commerce conversion rates increase 31% with localized payment methods” get pulled into answers verbatim. Sentences like “localization is important for global growth” get filtered out as marketing fluff.
The fix is structural. Replace adjectives with numbers. Replace generalizations with specific examples. Replace narrative buildup with TL;DR tables and definitions. AI search rewards documents that look like Wikipedia, not like Medium.
3. Mentions across the open web
This is the hardest one to fake. AI systems give weight to brands that appear repeatedly across independent sources — review sites, comparison articles, forums, industry reports. If your brand only appears on your own domain, AI search has no corroboration. It treats you as one source, not a category leader.
Building those external mentions is slower than SEO, but the half-life is longer. A single Reddit thread about your product can be cited by ChatGPT for years.
4. Proper hreflang and language signals
AI crawlers respect hreflang the same way Google does — when they can find it. If your French page is missing the hreflang tag pointing to your English original (and vice versa), AI search treats them as duplicate content from a confused site. It picks one to trust and ignores the other.
Most translation tools handle this automatically. If you’re translating manually or using a setup that doesn’t auto-generate hreflang, this is the first thing to fix.
The practical playbook
Here’s how to actually move the needle, in the order that matters.
Step 1: Audit which queries you’re invisible for
Open Perplexity and ChatGPT. Run 20 buying-intent queries about your product category — in every language you serve. Note which competitors get cited and which don’t. This is your baseline.
Don’t rely on automated tools for this yet; the AI search visibility tracking space is still maturing and most reports are unreliable. Manual checks in your top three languages will tell you more in an afternoon than any dashboard.
Step 2: Build language-native content, not translated marketing
Translation that reads like translation gets ignored. AI systems are surprisingly good at detecting machine-translated content with no editorial pass — and they downrank it relative to native-feeling content.
This doesn’t mean you need a human translator for every page. It means your translation pipeline needs at least one of three things: glossary control, native-speaker review on top-performing pages, or AI translation with locale-specific prompt engineering. Pick the one that fits your team.
Step 3: Restructure your content for extraction
Take your top 10 blog posts and rewrite them for AI extractability. That means:
- Lead each section with a one-sentence claim, then expand.
- Use tables, lists, and side-by-side comparisons wherever possible.
- Define every technical term the first time it appears.
- Replace adjectives with numbers.
- Add a “key takeaways” block at the top of every article.
The same content, restructured, can move from never-cited to frequently-cited within 60 days. We’ve seen it happen on our own blog and on three customer sites.
Step 4: Earn citations off-domain
Pitch comparison articles. Submit your tool to category review sites in each of your target languages. Run a small data study (translation pricing benchmarks, localization survey results, anything) and pitch it to industry publications. Every external citation that mentions your brand by name strengthens your AI-search authority.
This is where most companies stop, because it’s slower and harder to measure. It’s also where the biggest gains live.
Step 5: Implement technical signals correctly
Hreflang tags must be self-referencing and reciprocal. Every translated page should declare itself, every other language version, and the original. A common mistake: declaring hreflang only on the English page but not on the translations. AI crawlers and Google both treat this as broken.
Server-side translation beats client-side translation for AI visibility. Client-side approaches (where translations load via JavaScript after the page renders) are inconsistently crawled. ConveyThis and a few other tools render translations server-side, which means AI crawlers see the translated content the same way a human does.
What this looks like in 90 days
If you do all five steps consistently, here’s a realistic timeline based on what we’ve seen across customer accounts.
| Timeline | What you should expect to see | What’s still lagging |
|---|---|---|
| Days 1–30 | Foundational hreflang and structure fixes complete. Initial content restructure on top 10 pages. First external mentions secured. | AI search citations still rare. Most signals haven’t been picked up yet. |
| Days 31–60 | Restructured content gets indexed by AI systems. First citations appear in Perplexity for long-tail queries. | ChatGPT lags Perplexity by ~30 days for most brands. Gemini lags by ~60. |
| Days 61–90 | Citation frequency rises across all major AI search systems. Mid-funnel queries start surfacing your brand. Translated pages begin earning citations in their target languages. | Top-of-funnel category queries (“best translation software”) still take 6+ months to crack. |
These numbers aren’t promises — they’re the median we’ve observed across roughly 40 mid-market customers who applied this playbook seriously. Some moved faster. A few moved slower because their underlying content was too thin to support extraction, no matter how well it was structured.
The bigger shift
AI search visibility isn’t a separate channel. It’s becoming the discovery layer for everything else. Every quarter, more buying decisions start with an AI query and end with a click that your traditional analytics will misattribute to direct traffic or branded search.
If you treat it as an SEO sub-problem, you’ll under-invest. If you treat it as the new front door for your multilingual website, you’ll outpace every competitor still optimizing for blue links.
The brands that move first will become the default citations in their categories for years. The ones that wait will spend the next decade trying to dislodge them.
About the author
Alex Buran, Founder of ConveyThis
Alex has spent the last decade building infrastructure for multilingual websites. He writes about localization, AI search, and the technical side of going global.
LinkedIn: linkedin.com/in/alexburan
Want to know which AI queries your brand is invisible for? ConveyThis offers a free GEO (Generative Engine Optimization) audit that maps your visibility across ChatGPT, Perplexity, and Gemini in every language you serve. Book your free GEO audit
Summarize this post with:
Ready to Get Started?
Translation, far more than just knowing languages, is a complex process.
By following our tips and using ConveyThis, your translated pages will resonate with your audience, feeling native to the target language.
While it demands effort, the result is rewarding. If you’re translating a website, ConveyThis can save you hours with automated machine translation.
Try ConveyThis free for 3 days!
CONVEYTHIS