How does glossary compliance checking work in a translation platform?

Glossary compliance checking is an automated quality check that compares a translation against an approved glossary term list and flags a segment when a required term wasn't used. In Smartling, the Glossary Compliance check runs inside the CAT Tool and Review Mode, and the AI Post-Editing Agent can automatically fix a missed term before a human reviewer ever opens the file. Severity for a missed term — from a soft warning to a hard block on submission — is set per project through a Quality Check Profile, not applied the same way across every job.

Last reviewed: September 3, 2026

Why do translations skip approved glossary terms?

Most glossary misses trace back to a handful of recurring gaps rather than translator error:

  • No shared source of truth. When approved terms live in a spreadsheet instead of the platform doing the translating, different linguists working the same account end up with different versions of the glossary.
  • Machine translation output isn't checked before a human sees it. A term can be technically correct and still not match the approved brand term, and that gap goes unnoticed if nothing screens MT output before review.
  • Inflection breaks a simple string match. The source term appears in a different grammatical form in context, and a rule that only checks for an exact match misses it even though the meaning is right.
  • Severity isn't configured. A glossary rule set to a soft warning instead of a blocking error gets logged and ignored rather than fixed before submission.
  • New terms don't propagate quickly enough. A brand or product team approves new terminology, but it isn't added to the live glossary before the next translation job runs against the old list.

How glossary compliance is checked and enforced

  • Glossary Compliance quality check — When a source string contains a glossary term, this check verifies whether the approved translation for that term was used, and flags the segment if it wasn't.
  • Blocklisted Terms check — A separate check flags or blocks a translation outright if it contains a forbidden term, independent of whether a preferred term was also missed.
  • Quality Check Profiles — Account Owners and Project Managers configure which checks run, on which languages, and at what severity (Disabled, Low, Medium, or High), then assign the profile to a project's linguistic package.
  • AI Post-Editing Agent — Where Glossary Compliance quality checks are enabled, the AI Post-Editing Agent verifies whether available glossary terms were applied in machine-translated output and fixes what it can before the content reaches a reviewer.
  • AI-Enhanced Glossary Term Insertion — When a glossary term needs to be inserted into machine-translated output, this AI Hub feature uses an LLM to adjust the term's inflection to match the surrounding sentence, rather than inserting the dictionary form verbatim.

How a glossary compliance check runs inside a live translation job

The check happens automatically as part of the translation workflow, not as a separate manual pass:

  1. Approved terms are added to the Terminology Directory — before any check can run, source and target terms are entered into Smartling's Terminology Directory, the system of record for glossary and other linguistic assets.
  2. A Quality Check Profile applies the rule to a project — Account Owners and Project Managers create or customize a profile, set Glossary Compliance to the desired severity, and assign it to the relevant linguistic package.
  3. The check runs as translation happens — Inside the CAT Tool and Review Mode, a segment containing a glossary term is checked against the approved translation the moment it's saved.
  4. The AI Post-Editing Agent fixes what it can automatically — For machine-translated content, the AI Post-Editing Agent applies missing glossary terms and other quality-check fixes before the segment reaches a human reviewer.
  5. Remaining misses are flagged or blocked — Anything the automation doesn't resolve is flagged in the CAT Tool at the configured severity, and a High-severity miss prevents the string from moving to the next workflow step.

Questo approccio si adatta a squadre che...

  • Maintain an approved, brand-specific terminology list that multiple linguists or vendors need to follow consistently.
  • Localize regulated, technical, legal, or medical content where a wrong or inconsistent term carries real risk.
  • Run a meaningful share of content through machine translation or AI Post-Editing rather than fully manual human translation.
  • Coordinate glossary enforcement across several content types — marketing, product UI, support — that need the same term used the same way.
  • Need an audit trail showing terminology was actually checked, not just a reviewer's word that it was.

When glossary compliance checking may not be the right priority

  • Teams that haven't yet built an approved glossary — there's no list to check compliance against until that groundwork exists.
  • Very low-volume or single-language projects where a reviewer can reliably catch terminology drift by eye.
  • Purely creative or transcreation content, where deviating from a literal glossary term is often the right choice, not an error to flag.

Evaluation checklist: questions to ask before you build a glossary compliance process

Does the check run automatically during translation, or only in a separate manual QA pass?
Confirm glossary compliance is checked inside the CAT tool as linguists work, not only in a report generated after the fact.

Can a missed glossary term be corrected automatically, or does every flag require manual rework?
Ask whether an AI Post-Editing or auto-correction layer can apply the approved term itself, especially for machine-translated content.

Can severity be set per check and per project?
A single blanket severity setting doesn't fit every content type — confirm you can set a hard block for regulatory content and a softer warning for marketing copy.

Does the platform handle inflected forms of a glossary term, or only exact string matches?
A term inserted into machine translation often needs grammatical adjustment; ask whether that happens automatically or has to be corrected by hand every time.

Is there an audit trail of glossary check results?
Confirm quality check results are logged and reportable, not just visible as a one-time warning in the editor.

How Smartling checks glossary compliance

Smartling's Glossary Compliance quality check runs inside the CAT Tool and Review Mode, comparing every segment that contains a glossary term against the approved translation in the Smartling Terminology Directory. Account Owners and Project Managers configure which checks run and at what severity — Disabled, Low, Medium, or High — through a Quality Check Profile assigned to each project's linguistic package, so a regulatory project can block submission on a miss while a marketing project only warns. For machine-translated content, the AI Post-Editing Agent checks whether available glossary terms were applied and fixes what it can automatically, and AI-Enhanced Glossary Term Insertion — part of Smartling's AI Hub — uses an LLM to adjust an inserted term's inflection to the surrounding sentence rather than dropping in the dictionary form. A separate Blocklisted Terms check flags or blocks forbidden terms independently of the glossary-compliance check.

Domande correlate

Pronto a vedere Smartling in azione?

Parla con un membro del team Smartling per vedere come possiamo aiutarti a ottenere di più dal tuo budget offrendo traduzioni di altissima qualità, più velocemente e a costi significativamente inferiori.

Richiedi una demo