At Smartling, we’ve seen over and over again that translation memory (TM) leverage is the #1 driver that impacts high-quality translation outcomes. Ryan Soos, Sr. Customer Success Manager, discusses how TM leverage saves TMS users money, and why TM leverage can help translators create translations more quickly and with higher consistency.
What is Translation Memory?
Translation memory (TM) is a database of previously translated content that brands can draw from to translate similar strings of words. Since brands often repurpose phrases and words across different assets, rather than spending time and money to re-translate the same strings, they can work with their translators to insert translations from translation memory. It also enforces consistency of language throughout a brand's assets. A win for translators and their clients alike.
Translators can use translation memory based on how closely it matches the source content of their current project. Usually this is represented by a percentage, whether it's a 90% match, an 80% match, and so on.
Things to consider when using Translation Memory
Translation memory can be a great time and cost saver, but it's important to consdier whether or not your TM is of reliable quality. Maybe in previous projects, your translations weren't great. You may want then want to add a penalty to that TM so that your translators know the "match" is relatively low.
You also want to make sure your TM matches the tone of your particular project. For example, if your TM is based on translations of legal content, the stored translations may not make sense for branded marketing content.
For companies that have been translating for a while, they're likely to have multiple TMs to choose from. It's important they set within their translation management system (TMS) which TM to pull stored translations from first to ensure the best quality possible.
The more brands translate and store translations in their translation memory, the more their database grows and the more money they can save over time.