The most direct way of lowering translation cost is not to bargain over the unit rate but to reduce the number of words translated. This is the only method that does not give up quality.
First, what is really needed is determined
Translating the whole of a file is more often than not a habit. When it is asked which sections will be read by the counterparty, the list shortens markedly.
Annexes are assessed separately
Some of the annexes included for reference can stay in their original language. If the counterparty does not ask for them, translating them serves no purpose other than enlarging the volume.
Repeated sections are marked
It is normal for the same text to occur in more than one document. When these sections are translated once and reused, the total volume falls.
Text inside images is a separate decision
The wording inside diagrams and charts requires extra preparation. Whether it is really necessary is asked separately.
Old content is reviewed
Translating sections that are out of date means spreading obsolete information in one more language. The source text is cleaned up first.
A summary may be enough
For some documents what the counterparty needs is not the full text but an accurate summary. This is a different type of service, and it lowers the volume markedly.
The decision is put to the side that will use the text
It is not the person making the request but the person who will read the text who knows which sections are needed. If the two are different, the question must be put to the right address.
Counter-view: incomplete translation produces risk
That a section left untranslated may be needed later is a real risk. For this reason the decision is not made arbitrarily: it is written down which sections were not translated, and the document is archived with this information. In this way the gap stays visible.
Staged translation is an option
The critical sections are translated first, and the rest are added if the need arises. This manages both the volume and the risk.
The source text can be shortened
Some documents contain needless repetition and filler sections. Simplifying the source produces a gain in every language.
No decision is made before the volume is measured
An assessment of "a small file" made by eye is wrong more often than not. Once a count has been made, it can be seen which section is the heavy one.
The maintenance burden falls as well
Every text that is translated is translated again when it is updated. Unnecessary translation is not a one-off burden but a recurring one.
Reducing volume is not a decision about quality
The quality of the sections that are translated does not fall; simply fewer sections are translated. This distinction must be discussed openly.
The gain grows over time
Once scope discipline has been established, it is applied on every job and the gain accumulates throughout the year.
The decision is made once and applied on every job
Once it has been decided which sections of which type of document will be translated, this question is not asked again in later requests. Scope discipline thus ceases to be a decision made job by job and settles into the flow itself; the gain, too, repeats itself on every job of its own accord.
Takeaway
Reducing volume is the only way of lowering cost that does not touch quality. Determining what is really needed, assessing the annexes separately, simplifying the source and recording in writing the sections that are not translated; when these four are applied, the same job runs with a smaller volume.