A Translation Memory (TM) stores every segment your team has already translated — source text paired with its translation — so future projects can reuse that work instead of retranslating it from scratch. This article explains what a TM is, the settings that control how it matches and behaves, and how to connect one to your Organizational Units and projects.
​

💡 Who is this for? This guide is for Account Admins and Project Managers who need to set up and manage Translation Memories across their organization within the wxrks platform.

Every time a segment is confirmed — in the wxrks Editor, or added directly — it can be saved into a TM as a source/target pair. The next time a similar or identical segment appears, in any project the TM is connected to, the Editor can suggest that stored translation instead of starting from zero. This is what drives leverage: less retranslation, more consistency across projects, and lower cost as a TM grows.

A TM is never tied to just one project — the same TM can be linked to as many Organizational Units and projects as make sense for your account, and it keeps accumulating content across all of them.

A few terms come up throughout this article and the rest of the TM settings screens:

From the Context menu, click Translation Memories to see every TM already created, and use Add TM to start a new one.

Creating a TM only takes a name, one Organization, one source language, and one or more target languages — and once it's created, that Organization can never be changed. See How to create a Translation Memory for the full field-by-field walkthrough.

Opening a TM shows its Organization, the Organizational Units it's linked to, its total stored segments, source/target languages, and last import date, alongside three actions:

Import loads content into the TM from .tmx or .xlsx files — up to 60 files in one batch, queued and processed automatically. Segments can also be added one at a time, or confirmed straight from the Editor; see How to add new segments to Translation Memories.

The Actions menu holds the rest of the TM's own management options:

Edit changes the TM's name and target languages. Export downloads the TM's content as a .tmx file — see How to export the Translation Memory for the full steps. Clear removes every stored segment without deleting the TM itself, and cannot be undone. Delete removes the TM entirely, but only once it's no longer linked to any Organizational Unit.
​

If TM matches aren't showing up as expected in the Editor even though the content looks right, an Account Admin can force a resync from the TM Troubleshooting button next to Import.

Every TM has two settings that apply the same way everywhere it's used, opened from its gear icon wherever the TM appears in an Organizational Unit's or project's Context tab:

CODE/UoM ignores differences in part numbers and units of measure when matching, so segments like "224mA" and "448mA" still count as the same TM match. Penalty deducts a percentage from every match this TM returns — a 100% match with a 5% penalty shows as 95% — so reused content competes fairly against a fresh translation instead of always winning by default.

This same modal also has a Settings by Workflow section for overriding a TM's Read/Write behavior on individual workflow steps (letting a TM contribute matches during Translation but stay read-only during Proofreading, for example). That's covered in full in Configure Granular TM Workflow Settings by Org Unit and Project.

A TM only becomes usable in a project once it's linked to that project's Organizational Unit. Open the Unit's Context tab, and use Add in the Translation Memory table to link any TM that belongs to the Unit's Organization.

Each linked TM gets its own Default, Read, and Write toggles and Threshold here — these become the TM's starting settings for every new project in this Unit. See How to create a Translation Memory for the step-by-step of the linking flow itself.
​

When a TM is linked to an Organizational Unit, it becomes available to every project inside that Unit. A project's own Context tab lists the TMs it's actually using — click Add TM there to link one manually, or rely on any TM marked Default at the Organizational Unit level to be added automatically when the project is created.

A project's Read, Write, and Threshold values start out copied from the Organizational Unit's own settings for that TM, then become independent — changing them here doesn't affect the Organizational Unit, and a later change at the Organizational Unit level doesn't retroactively update projects that already have the TM linked. See the Project Overview article for how this Context tab fits in alongside the rest of a project's settings, and Configure Granular TM Workflow Settings by Org Unit and Project for per-workflow overrides at either level.

Once a TM has content, its own Search panel supports full text, fuzzy, and exact/regex search across its stored segments — see How to use the Translation Memory search for the full breakdown of each search mode.