Software Localization - some details in terms of how the process work - Part 2

In the previous post on this topic (Some details of software localization), I described the localization process at a high level, with not too many execution details. In this post, I will continue at the same high level, still continuing to explain the process of localization and its relation to the product schedule.
When you consider a product schedule, there are many milestones that you need to meet, with one of the milestones being a state whereby you freeze all the UI related content, specifically strings (these could be text on the dialogs, could be error messages, popups that appear when the mouse hovers over certain parts of the screen, etc). The milestone where the strings need to be frozen can be a single milestone in the second half of the schedule, or it could as part of periodic milestones where individual features are frozen in terms of their UI.
Once the strings are frozen, they can be pulled out (as described in the first post in this series), and then sent to the appropriate language vendors for translation. Once the translation happens and these translations are sent back, the translations can be incorporated in the product code (in a resource file that contains all the strings for different languages). The net result is, when the product is launched in a specific language, the application picks up the relevant language strings and shows them.