More flexibility in your quality checks: here’s what’s new in XTM Cloud 26.4

Over the past few years, localization workflows have been changing. Depending on the needs of your business and your content, translations can now come from many different sources, whether that's a human linguist, an MT engine, or an LLM. This gives teams more options than ever, but it also raises an important question: how can you be confident that every translation, wherever it came from, has been checked for quality against the same standard?

For many teams, the natural response is to review everything, just to be sure nothing slips through. It's an understandable approach, but as content volumes grow and timelines get shorter, it becomes harder to sustain.

This is where XTM Cloud 26.4 can help. With this release, the Translation Quality Index (TQI) becomes a single, consistent quality layer across every translation source. It’s now available as an auto-step, a fully automated step you can add to your workflows, so quality checks become part of your process.

Here's what's new:

• One score, any source: TQI can now score translations whether they come from TM, MT, an LLM, or a linguist. Every segment is measured against your quality standards, so you can compare quality across sources.
• Quality built into the workflow: XTM Cloud Administrators can create or edit automatic workflow steps with TQI enabled, and include those steps in workflow definitions. Quality checks become a standard part of how work moves through the platform.
• Ready for new and existing projects: Project Managers can choose workflow definitions that include TQI auto-steps when creating a project, or add TQI auto-steps to existing projects directly in Project Editor.
  

A closer look at how TQI evaluates quality

TQI gives every translation a score based on how well it meets your quality standards, but there's a lot more behind that number than a single check. It brings together three types of evaluation:

• Translation complexity: Shows how difficult a string is to translate and how much attention it's likely to need. TQI looks at how complex or ambiguous the source text is, and compares how several different AI models would translate it. If the models agree, the string is likely straightforward. If they don't, or the text is open to more than one meaning, TQI highlights it for a closer look.
• Structural integrity: This verifies the technical side of a translation, such as terminology, tags, placeholders, numbers, and formatting.
• Semantic quality: Based on MQM (Multidimensional Quality Metrics), a widely recognized industry framework that looks at areas like accuracy, grammar, style, and audience appropriateness.

Together, they give a far more complete picture of quality than either approach could on its own, and it all runs fully automatically.

What makes TQI especially powerful is that it knows your setup. It checks every translation against the approved terms in your termbase, so product names, key phrases, and your corporate terminology are used consistently across languages. It also applies the rules in your style guide, from tone and formality to locale conventions, so content is evaluated against how your brand actually communicates, not a generic idea of what "good" looks like.

From a score to better decisions

• See what's wrong and why: Alongside every score, TQI provides a structured issue analysis, with each issue categorized and given a severity level. Reviewers can go straight to what matters instead of working out what caused a low score.
• Compare every source on equal terms: Because TM, MT, LLM, and human translation are all evaluated the same way, you no longer have to assume a translation is good because of where it came from. Even a 100% TM match might use outdated terminology or not fit its new context, and TQI helps you catch that.
• Spot patterns over time: Because issues are categorized consistently, quality becomes structured data. Teams can find recurring problems by language, content type, workflow, or source, and act on them, from updating their termbase and style guide to adjusting workflows or translation strategies.

Flexible enough for projects of any size

Because TQI runs as a batch task, you decide exactly where and when it's applied. Working on a large project with 20 languages and 200 files, but only need to check one document in one language? You can run TQI on just that. Want to assess your source content before localization begins, so potential issues are caught at the very start? That's possible too. With this level of control, you can fit quality checks around the way each project works, rather than changing your projects to fit the checks.

Focusing effort where it matters most

Content volumes keep growing, and so does the variety of ways that content gets translated. Reviewing every segment manually isn't always practical, and spot checks do not give the full picture. By scoring every translation automatically, including TM matches and human translations, TQI helps teams see where human expertise can have the greatest impact, reducing post-editing overhead and making better use of their localization budget.