How do Translation Memories work?

Translation Memory draws on previously approved translations as a guide to ensure consistency in your brand and tone of voice, building a record of your content over time across languages and markets. By reusing previously approved translations, translators can focus on new and updated content. This helps them work more efficiently so they can deliver faster and cheaper for you.

"Segments" and "matches"

The content you've sent us is filtered to produce "segments". While there are exceptions (typically software content), a segment is usually considered to be a full sentence.

These segments are then matched against your previously translated content to identify if there are any similarities ("matches") to be found between them. 

Potential Translation Memory matches range from:

• In-Context Exact Match - a segment that has an exact match, and is preceded and followed by the same text as previously seen. This requires no change.
• Leveraged/100% Match - the segment is the same, however, the segments before or after it are different. This segment may require adaptation due to the different context surrounding it, meaning that it needs to be reviewed by the translator for changes.
• "Fuzzy" Match - matches often ranging from 99% to 75% that indicate a similarity between your new content and the previous translation.
• No Match - a segment containing new text for which a match hasn’t been found, requiring a full translation.

Our Computer-Assisted Translation (CAT) tool also saves useful metadata with each segment our translators work with, such as who created it, who updated it, when it happened, how many times it has been re-used, etc.