Why translation is not research
The instinct is reasonable: you have a keyword list that works, you are entering a new market, so you translate the list. The problem is that a keyword is not a concept, it is the specific string a population types — and the mapping between concept and string is cultural, not linguistic.
Three ways it breaks. Regional vocabulary: Spain searches "móvil", Mexico searches "celular", and a translator given "mobile phone" will produce whichever their own dialect prefers. Borrowed English: in many markets the English term outranks the native one — German developers search "cloud hosting", not a translation of it. Different demand shape: the same product can be a high-volume commodity search in one country and a near-zero niche in another, so a page worth building in one market is not worth building in the next.