If you work with translated content regularly, you have probably heard the terms translation memory and glossary used interchangeably. They are both essential tools in any serious localization workflow, but they solve very different problems. Understanding those differences and knowing when to reach for each one will save you time, improve consistency, and reduce costly revisions.
What Is a Glossary?
A glossary is a curated list of term-to-term mappings in two or more languages. Each entry pairs a source term with its approved translation. Glossaries are typically short, focused, and managed by product, marketing, or localization leads who decide how specific words should appear across all translated content.
- Brand names that should never be translated or should follow a specific localization pattern
- Product feature names that need a consistent label in every language
- Technical jargon unique to your industry or domain
- Legal terms that require exact, pre-approved wording
For example, if your product has a feature called "Smart Compose," a glossary entry ensures every translator and every engine renders it as "Smart Compose" (untranslated) or as the single approved localized equivalent in each target language.
What Is Translation Memory?
Translation memory (TM) is a database that stores previously translated segments — typically full sentences or paragraphs — alongside their source text. When a new piece of content is submitted for translation, the system checks the TM for exact or fuzzy matches. If it finds a match, it reuses the stored translation instead of translating from scratch.
- Exact matches (100%) are reused automatically, saving both time and cost
- Fuzzy matches (typically 75-99%) are presented as suggestions that a translator or reviewer can accept, edit, or reject
- No match segments are translated fresh and then added to the TM for future reuse
Translation memory is especially powerful for content that is updated frequently, such as documentation, help articles, release notes, and software interfaces where large portions of text stay the same between versions.
Key Differences at a Glance
| Aspect | Glossary | Translation Memory |
|---|---|---|
| Scope | Individual terms or short phrases | Full sentences, paragraphs, or segments |
| Granularity | Word-level or phrase-level | Segment-level |
| Typical use | Enforcing brand vocabulary and terminology | Reusing past translations to speed up new projects |
| Storage | Small, curated list (tens to hundreds of entries) | Large database (thousands to millions of segments) |
| Who manages it | Product, marketing, or terminology managers | Localization engineers or translation platforms |
| Growth | Manual — terms are added deliberately | Automatic — grows with every completed translation |
When to Use a Glossary
Reach for a glossary when consistency of specific terms is the priority:
- Brand terms and product names that must be rendered identically everywhere
- Technical jargon where a wrong synonym could confuse users or violate compliance
- Legal and regulatory terms that have pre-approved translations
- New product launches where terminology has not yet appeared in any TM
When to Use Translation Memory
TM shines when you are translating large volumes of content with significant overlap:
- Documentation updates where 80% of the content is unchanged between versions
- Software localization with recurring UI strings across releases
- Large website migrations where many pages share common boilerplate
- Ongoing content programs that produce similar articles or support pages regularly
How They Work Together
The real power comes from combining both tools. A glossary enforces that key terms are always rendered correctly, while TM handles the heavy lifting of reusing full segments. Here is a typical workflow:
- The glossary is applied first, locking in approved terminology before translation begins
- The TM is checked next, pulling in any matching segments from previous projects
- Only truly new content is translated fresh, with glossary terms guiding the engine or translator
- Completed translations are stored back into the TM for future reuse
Best Practices
- Keep glossaries small and focused. A glossary with thousands of entries becomes hard to maintain and can conflict with natural phrasing. Aim for 50-200 high-impact terms per language pair.
- Review TM regularly. Outdated or low-quality translations in your TM will propagate errors. Schedule quarterly reviews to purge or update stale segments.
- Combine both for maximum efficiency. Use the glossary to set the rules and the TM to scale. Together, they reduce costs, improve consistency, and accelerate turnaround.
- Version your glossary. Track changes over time so you can audit when a term was added, changed, or removed — especially important in regulated industries.
How Adara Helps
Adara Translate includes a built-in glossary feature that lets you define term pairs and apply them automatically across all supported engines — DeepL, Google, Microsoft, GPT, and Gemini. When you create a glossary in Adara, every translation request respects your terminology rules, ensuring consistent output no matter which engine you choose. This is especially valuable for teams that need brand-consistent translations at scale without managing a separate terminology tool. For a step-by-step localization plan, see our website localization checklist. And for tips on keeping API costs down while maintaining quality, read how to reduce translation API costs.
