When you open a segment in the wxrks Editor, every linked Translation Memory (TM) is checked for a matching translation — but finding a match doesn't automatically mean it gets typed into the segment for you. wxrks only writes a TM match directly into a segment when several separate settings all line up; when even one doesn't, the match still shows up as a suggestion, but someone has to apply it by hand. This article walks through the real reasons a TM match won't auto-fill and exactly which setting controls each one.
​

💡 Who is this for? This guide is for Vendors and Project Managers who need to diagnose why a Translation Memory match isn't auto-filling a segment within the wxrks platform.

A TM match can reach a segment two different ways. It can appear in the Editor's Suggestions panel for you to review and insert yourself — this happens whenever a stored match exists at all, regardless of settings. Or it can be auto-filled: written directly into the target segment by an auto-translation pass (also called Pre-Translation) that includes Translation Memory as one of its sources. The reasons below are all about that second path — why a match that looks like it should qualify still isn't the one that ends up in the segment.
​

For what Threshold, Penalty, and the difference between a 100% and 101% match actually mean, see All about Translation Memories — this article assumes those basics and focuses only on why auto-fill specifically doesn't happen.

Auto-translation (Pre-Translation) can pull from up to three sources — Context Sensitive, Machine Translation, and Translation Memory — each its own toggle on the project's Translation Settings tab. If Translation Memory itself isn't switched on there, no TM match will ever auto-fill, no matter how high it scores. It'll only ever show up as a suggestion.
​

Turn on Translation Memory under Auto Translate From, then run Pre-Translation again — see Project's Translation Settings for exactly where this toggle lives.

Once Translation Memory is enabled as a source, a threshold slider appears: only matches at or above that percentage actually get auto-filled. Anything lower stays a suggestion on purpose, so a linguist reviews it instead of the platform silently shipping a rough substitution.
​

A less obvious version of the same rule: if a segment has more than one Translation Memory entry that would otherwise each be a 100% match, wxrks treats that as ambiguous — it doesn't guess which one you want — and reduces the score by the account's Penalty for multiple 100% match setting instead. A match that looked perfect can drop below the threshold this way and stay a suggestion, even though nothing about the match itself changed.

If a project also has Context Sensitive (AI) translation turned on, it's the default winner for a segment even when a strong TM match exists — Context Sensitive already blends Translation Memory into its own suggestion internally, but that's not the same as the raw TM match itself being the thing written into the segment. To let a Translation Memory match win outright over Context Sensitive, a Project Manager or Account Admin has to turn on Prioritize TM matches over Context Sensitive translation and set a Minimum TM match threshold. Until that's on, a good TM match can exist and score well without ever being the one that lands in the segment.

A few settings narrow which segments an auto-translation pass is even allowed to touch:
​

When a stored TM entry carries formatting tags that don't line up exactly with the current source segment's tags, wxrks raises a tag-mismatch warning instead of auto-filling — inserting a match with the wrong tag structure risks corrupting the file's formatting on export, so it holds back rather than guessing. Turning on Ignore the Tags mismatch Warnings lets the match through anyway; when you do, check the tags on those segments manually afterward, since they may still need correcting by hand.

A project's Translation Settings only affect segments translated from that point forward. Turning on Translation Memory, raising a threshold, or enabling Ignore the Tags mismatch Warnings today doesn't reach back and re-apply to segments that were already auto-translated, confirmed, or locked under the old settings — those stay exactly as they are until Pre-Translation runs again or someone edits them by hand.
​

The same rule applies one level up: Unlock options and Mismatch Warnings are set once on the Organizational Unit's Translation Settings template and only get copied into a project at the moment that project is created. Fixing them on the Organizational Unit doesn't change anything for a project that already exists.

A match sitting in the Suggestions panel instead of the segment isn't a bug — click it to insert it yourself, the same as any other suggestion, before confirming the segment. If you're a Vendor and don't have access to change threshold, priority, unlock, or tag-mismatch settings yourself, the fastest path is usually flagging the pattern to your Project Manager or Account Admin — per the reasons above, most of these are one setting away from being fixed for the rest of the project.