Maintaining multilingual content: titles, descriptions and proofreading
The usual problem on multilingual sites is not translation quality but that nobody looks regularly. Chinese titles leaking into English pages, stale translations and descriptions of removed features quietly erode trust.
How to do it
Start with an automated scan
Scan titles and descriptions for characters that should not appear in that language, such as Chinese characters in an English title. Treat the output as a backlog, not a one-off cleanup.
Review in three passes: language, meaning, currency
Language: does it read naturally in the target language. Meaning: does it still match the body. Currency: are the features and prices mentioned still valid. Do these separately so you do not stop at grammar.
Fix titles and descriptions first
Titles and descriptions are entry points, so fix them first; changing the body costs more and is only needed when the meaning is wrong. Log each fix in the handoff so nobody repeats it.
Make translations read like native writing
Ask for a rewrite rather than a polish, then read it once: Chinese word order, repeated particles, over-nominalisation are the usual tells.
Set a review rhythm
After each publish, spot-check three older items in all languages; once a month run a full scan of titles and descriptions and put the findings into the backlog.
Prompts used
FAQ
Should brand names be translated?
Usually keep the original and translate only the descriptive part, so readers still recognise it.
Must the translation match the Chinese exactly?
No, but facts, numbers and promises must match; tone and sentence structure should follow the target language.
How often should the full scan run?
Monthly is enough; quarterly tends to pile up if you publish often.