Ask ten managers how they judge the quality of their translations and you will hear ten different answers. Some rely on a bilingual colleague's opinion. Some wait for customer complaints. Others simply trust their supplier. None of these methods is wrong, but none of them gives a reliable picture, especially when a company translates large volumes into many languages.

Measuring translation quality properly is possible, and it does not require a research department.

Define what quality means for you

Quality is not a single thing. A legal contract needs absolute accuracy and consistent terminology. A marketing campaign needs persuasive, natural language. A user manual needs clarity and correct technical terms. Before measuring anything, decide which aspects matter most for each type of content.

Common dimensions include accuracy of meaning, terminology, fluency and style, grammar and spelling, formatting, and compliance with any client style guide. Weight them according to the content type.

Use error categories and severity levels

Professional quality frameworks classify errors into categories such as mistranslation, omission, addition, terminology, grammar, punctuation and style. Each error also receives a severity level: minor, major or critical. A critical error changes the meaning in a way that could cause harm, legal risk or financial loss. A minor error is noticeable but does not affect understanding.

By counting weighted errors per thousand words, you get a score that can be compared across projects, suppliers and languages. This is far more useful than a vague impression that a text is "good" or "not quite right".

Sample rather than review everything

Reviewing every word of every translation is expensive. A practical alternative is sampling: select a representative portion of each project, for example 10 percent or a fixed number of words, and evaluate it in detail. Over time, the scores reveal patterns. One language may consistently score lower, or one content type may cause more terminology errors.

Standards provide a baseline

International standards describe what a professional translation process should look like. ISO 17100, for example, sets requirements for translator qualifications, revision by a second linguist and project management. Working with suppliers certified to such standards does not guarantee perfect results, but it ensures that basic quality steps are in place.

Use technology to catch the obvious

Automated quality checks catch many mechanical issues before a human reviewer ever sees the text: missing numbers, inconsistent terminology, untranslated segments, double spaces and tag errors. Larger organisations usually run these checks inside an enterprise translation management system, which also stores review scores, tracks trends over time and links feedback directly to the segments concerned.

Language-specific challenges

Some language pairs need special attention. A translation english to russian must handle grammatical cases, gender agreement and much longer sentences, which makes layout issues and agreement errors more common. Reviewers should know what to look for in each language rather than applying one checklist everywhere.

Official documents are another special case. For french document translation services used in visa or court proceedings, a single error in a name, date or number can lead to rejection. Here, the tolerance for errors is effectively zero, and quality checks should focus on data accuracy above all else.

Close the feedback loop

Measuring quality is only useful if the results lead to improvements. Share review findings with translators, update glossaries and style guides, and discuss recurring issues with suppliers. Good translators welcome specific, fair feedback because it helps them meet expectations more consistently.

Involve in-country reviewers carefully

Many companies ask local sales teams or distributors to review translations. Their market knowledge is valuable, but their feedback can be subjective or inconsistent. Give them clear instructions, a simple form and the same error categories used by professional reviewers. Distinguish between genuine errors and personal preferences.

Track the business impact

Ultimately, translation quality matters because of its effect on customers and operations. Track metrics such as support tickets related to unclear documentation, conversion rates on localised pages, and customer reviews that mention language. These indicators connect quality scores to real outcomes.

Start with better source texts

Many translation errors begin in the original. Ambiguous sentences, inconsistent terms and missing context force translators to guess. Improving source content, with clear writing guidelines and a terminology list, often raises translation quality in every language at once, and it costs far less than fixing the same problem ten times downstream.

A simple framework to start

  • Define quality priorities for each content type.
  • Adopt error categories and severity levels.
  • Review a representative sample of each project.
  • Use automated checks for mechanical errors.
  • Share feedback and update resources regularly.
  • Link quality scores to business results.

Quality as a shared goal

Translation quality improves fastest when clients and suppliers treat it as a shared responsibility. Clear expectations, good source texts, useful reference materials and structured feedback create the conditions for excellent translations. Measurement is not about catching translators out. It is about building a process that delivers reliable results, project after project.