The Black Art of Estimating Translations

Almost all the source content that we spit out of or WCMS goes out to our vendor for translation into 32 languages. They constantly demand accurate estimates of word count changes for each language (so they can make sure they have adequate staff in each language), but can never give us a good method for determining this estimate.

For changes to content that has been previously translated, the best we can do is to count all the words in each affected sentence, add up all the changes, then reduce the count by about %20 to account for translation memory savings. But sometimes translation memory automatically takes care of %75 of the changes and sometime almost none at all. For us, there is just no way of knowing which it will be until it runs through their system.

If anyone has a better way, I’d love to hear it!