Brands serving both Thai customers and international visitors usually do one language well and translate the other. Search notices, and so do the AI engines. Translated content reads like translated content, and it rarely matches how people actually search in the second language.
Search intent is not the same in both languages
A Thai customer and an English-speaking visitor looking for the same thing often phrase it completely differently, use different reference points, and expect different proof. Keyword research done in one language and translated into the other misses the queries that actually convert. Each language needs its own research, against its own results page.
The technical layer has to agree with you
Hreflang, clean language paths, and consistent structured data tell both search and AI engines which version to serve to whom. Get this wrong and the two versions compete with each other, or the wrong one surfaces for the wrong audience. Get it right and each language compounds on its own terms.
AI engines read both, and judge both
When someone asks an AI assistant in Thai for the best in your category, it leans on Thai-language sources and signals. A brand with strong English content and thin, translated Thai content shows up confident in one answer and absent in the other. Native content in both languages is what earns citations in both.
The fix is not twice the work for its own sake. It is treating each language as a first-class audience: research, write, and optimise natively in both, hold the technical layer steady underneath, and let each side build authority instead of leaning on the other. That is how a bilingual brand stops leaking half its market.