On an afternoon last week we held an informal internal session in the office. There were no slides and no sign-in sheet. Several colleagues wrote the three questions clients ask most often on the whiteboard: how is this translated, why is the name different from last time, and can a crooked scan still be used? After looking through two pages of notes, Wang said it made more sense to write a checklist so that whoever met the question next could add a line.
It was a small thing, but it grew out of real material. That week we had several sets of shipping documents and bank papers on the desk, and the questions kept coming while the answers often required digging out an old draft. Everyone agreed that an answer written on paper was more reliable than one left in a chat thread.

Question one: how is this translated
When a client asks how a sentence is translated, what they usually mean is which wording their own trade uses. The document on the table that day was a shipping description mixing product names and model numbers with a few lines that looked as if they had been added casually. We first separated the fields that had to be matched word for word from the lines where the sense was enough, and only then decided how much effort each deserved.
One line had appeared in two different forms in an old draft, and nobody could say which was the later revision. Rather than force a decision, we left a row in the table recording where each form appeared and noting that we would ask the client which one to use next time. Tang said this was exactly the place where "close enough" did not do: one different letter becomes another matter at the port.
We also added a column headed "what this trade calls it". The same English word can carry one name in a shipping document and another in a product leaflet. Without that column, the next person simply follows the old draft and repeats the mistake. Wang said this was the most useful part of the table, because it records not an answer but the way to ask the question.
Question two: why is the name different
Names are more tedious. A company name in two documents differed by one space; a client's name was spelled differently in a passport and an older certificate. It is tempting to call both a mistake on the other side, but some are leftovers from earlier records and some are simply habits of entry. The first rule we wrote down was not how to correct them but what order to follow: find out what the papers will be used for, then decide which version to follow and whether a note is needed.
That led naturally to bank statement translation and stamping. The account name on such papers often differs slightly from the one on a business licence, and the scan is usually stamped and slightly crooked. The rule we settled on that day was plain: check each field against the image first, then talk about wording. We only translate and apply our registered translation seal; whether a document is acceptable is for the receiving side to decide.
Question three: can a crooked scan be used
This one comes up almost every week. Some scans are stamped, some show only half a page, some have figures hidden in shadow. Hu pulled out an old draft as an example: one figure had been read as another, which took minutes to correct and two days to confirm. So another line went into the table: mark what you cannot read and ask, never guess.
There is no trick to it. Better one extra question than a "probably" in the translation. The checklist grew to a page and a half, with a few blank lines left for the next time.
Another small incident came up in the middle of the session. A colleague had placed two photos from a client side by side and noticed that the same page had been photographed twice, with one figure different. It was caught before delivery. Hu used it to add "compare versions before you start" to the table, right under "ask when you cannot read it".
After the list went on the wall
Before we finished, we copied the list onto a sheet by the wall and put another into the folder for document translation. No new service was added and no results were counted; we simply turned "thinking it through again every time" into "reading it off the list". These questions sound trivial, yet they are the ones most worth writing down in this line of work.
Document problems rarely appear out of nowhere; the same one comes back in a different form. Write it down and you save a phone call next time. The industry is talking about standardised fields in electronic documents and multilingual data; what we can do at our desk is get the names and fields on one sheet to agree. The industry piece and the translator piece in this batch go into more detail.
