Glossary Management: Consistent Brand Terminology Across Every Language

Terminology inconsistency is one of the most damaging — and most common — problems in product localization. Your product calls the main navigation item a "workspace." Your Spanish translation has it as "espacio de trabajo" in the dashboard, "área de trabajo" in the settings, and "proyecto" in the help docs. Users are confused. Support tickets spike. Trust erodes.

This isn't a translation quality problem. It's a terminology governance problem. And translation glossary management is the solution.

better-i18n's Glossary Management system gives you complete control over how your product's vocabulary is translated across every language, enforced automatically at the AI translation and review stages — so inconsistencies never make it to production.

How It Works

A glossary in better-i18n is a structured database of terms: the source language word or phrase, its approved translation in each target language, usage notes, context restrictions, and special rules like "never translate" or "always capitalize."

Adding Terms

Terms can be added manually through the glossary editor, imported in bulk from JSON files, or automatically extracted by the Context Crawler, which analyzes your live product or codebase to identify candidate terms. Each term supports:

Source term (with optional regex for inflected forms)
Target translations per language
Definition and usage notes
Part of speech (noun, verb, adjective)
Domain tag (UI, marketing, legal, technical)
Never-translate flag
Case sensitivity rules

Term Workflow: Draft to Approved

Every glossary term follows a structured lifecycle that ensures quality and accountability:

Draft — A new term is created (manually, via import, or proposed by the Context Crawler). Draft terms are visible to the team but not yet enforced in translation workflows.
Submitted — The term author or a team member submits the term for review. This signals that the term definition and translations are ready for evaluation.
Approved — A reviewer approves the term. Only approved terms are enforced during AI translation and highlighted in the review editor. This prevents half-finished or contested terms from affecting translations.
Rejected — A reviewer rejects the term with feedback. The author can revise and resubmit.

This workflow prevents the common problem of glossary pollution — where untested or disputed terms get enforced prematurely and cause more inconsistency than they solve.

Bulk Import and Export

For teams migrating from another tool or managing large glossaries across multiple systems, better-i18n supports comprehensive bulk operations:

JSON Import — Upload a JSON file with terms, translations, and metadata. The importer validates entries, deduplicates against existing terms, and reports conflicts before applying changes.
JSON Export — Export your entire glossary or a filtered subset as structured JSON. Use this for backup, migration to other tools, or feeding terminology into external workflows.
Clear — Remove all terms in a single operation when rebuilding a glossary from scratch. This is protected by a confirmation step to prevent accidental data loss.

These bulk operations make it practical to manage glossaries with thousands of terms without manual entry overhead.

DeepL Glossary Sync

For teams using DeepL as their machine translation provider, better-i18n offers automatic glossary synchronization:

One-click sync — Push your approved glossary terms to DeepL's glossary API. DeepL will then enforce your terminology in all machine translations processed through their service.
Bidirectional consistency — Your glossary becomes the single source of truth for both better-i18n's AI translation and DeepL's neural machine translation. No more maintaining separate term lists in two systems.
Language pair mapping — better-i18n automatically maps your multi-language glossary into the language-pair format DeepL requires, handling the conversion transparently.

This integration means your brand terminology consistency extends beyond better-i18n's own translation engine to any workflow that touches DeepL.

Enforcement During AI Translation

When AI translation runs, the full glossary of approved terms is injected into the translation prompt. The AI model is instructed to use approved terms exactly as specified. After translation, a consistency validator checks the output against glossary entries and flags any violations before the translation is surfaced for review.

Enforcement During Human Review

In better-i18n's review editor, glossary terms are highlighted inline. When a translator types a variant of a term, an autocomplete suggestion appears with the approved glossary translation. Terms used incorrectly are underlined in yellow, with a tooltip explaining the approved term and why it matters.

Never-Translate Rules

Some terms should never be translated — product names, brand names, technical identifiers, competitor names, and trademark terms. Mark these with the never-translate flag and they will be passed through unchanged in all AI and human translation workflows, regardless of language.

Key Benefits

1. Brand Terminology Consistency at Scale

When your product is used in 15 languages by thousands of users, every instance of an inconsistent term is a UX problem. Translation glossary management makes consistency automatic rather than aspirational. Once a term is defined and approved through the workflow, it is enforced everywhere — in AI translation, in human review, in DeepL output, and in export validation.

2. Faster Onboarding for New Translators

When a new translator joins your project — whether in-house, freelance, or from a translation agency — they immediately have access to the complete glossary. They don't need to read through previous translations to learn your terminology. The inline editor guides them in real time.

3. Reduced Review Cycles

Terminology corrections are one of the most common reasons translation batches go back for revision. When glossary enforcement catches these issues before review, the review cycle focuses on style and nuance rather than basic terminology compliance. Review time decreases significantly.

4. Automatic Glossary Extraction

Building a glossary from scratch is a significant undertaking. better-i18n's Context Crawler automates the most tedious part: it scans your live website using Firecrawl-powered analysis and parses your GitHub repositories to identify repeated terms and branded language. Proposed terms enter the glossary as drafts, ready for your team to review and approve. A glossary that would take weeks to build manually can be bootstrapped in hours.

5. Legal and Compliance Protection

For regulated industries, certain terms must be translated with precise legal equivalents. Pharmaceutical, financial, and legal products use the glossary to enforce these requirements, with audit trails showing that approved translations were used in production content.

6. Provider-Level Enforcement via DeepL Sync

Even when translations happen outside better-i18n's editor — through DeepL API calls in your CI pipeline or third-party integrations — your glossary is still enforced. The DeepL sync ensures your terminology rules travel with the translation engine, not just the translation tool.

Use Cases and Examples

Product UI Terminology

A project management SaaS has a core concept called a "Sprint." In German, the team has decided to keep it as "Sprint" (not translated). In French, the approved term is "Sprint" (same). In Japanese, it's rendered in katakana. The glossary stores all of these, with the English source marked as never-translate for English-language markets. Every AI translation and human review enforces this automatically.

Marketing and Brand Language

A brand's tagline includes the coined term "workstream." The glossary specifies approved translations in 12 languages, with the English term listed as a never-translate for all markets where the brand runs English-language campaigns. AI translation respects this — the term never gets "helpfully" rewritten.

Technical Documentation

A developer tools company has a glossary of 300 technical terms — API concepts, CLI commands, configuration keys, and error codes. None of these should ever be translated. The never-translate flag on all 300 terms means they pass through untouched regardless of which translator or AI model processes the documentation. The team exports the glossary as JSON monthly for their external documentation pipeline.

E-commerce Product Taxonomy

An e-commerce platform has precise category names that must match local market conventions. The glossary stores locale-specific category translations with context restrictions — "Jackets" translates differently when it appears in men's vs. women's clothing categories. Context-restricted entries handle this precisely.

Large-Scale Migration

A gaming company migrating from a spreadsheet-based glossary with 2,000 terms uses the JSON import to bring everything into better-i18n in one step. After import, they run the approval workflow to validate each term, then sync approved terms to DeepL for their localization pipeline.

How better-i18n Implements Glossary Management

better-i18n's glossary system is tightly integrated with every stage of the translation pipeline — it is not a standalone tool that translators have to remember to check.

Glossary as Context

When a translation job runs, the relevant subset of the glossary (approved terms that appear in the source strings) is automatically extracted and included in the AI prompt. This is more effective than providing the entire glossary, as it keeps the prompt focused and reduces token cost.

Validation Layer

A fast secondary model runs a consistency check on every AI translation output, comparing used terminology against the expected glossary translations. This happens in milliseconds and catches issues before they reach human reviewers.

Glossary Versioning

Glossaries evolve. Terms get renamed, rebranded, or refined. better-i18n versions every glossary change, allowing you to see what translation decisions were made under which glossary version. You can roll back a glossary change and trigger re-validation of affected translations.

Integration with Translation Memory

The Translation Memory and glossary work together: TM provides segment-level reuse, while the glossary enforces term-level accuracy within those segments. When a TM match is applied, it is still validated against the current glossary — so if the glossary changed since the TM entry was created, the inconsistency is flagged.

Comparison with Alternatives

Feature	better-i18n	Phrase (Memsource)	Lokalise	Spreadsheet
Term approval workflow	Draft > Submitted > Approved	Basic	Basic	No
Inline editor highlighting	Yes	Yes	Yes	No
AI prompt integration	Native	Plugin	Plugin	No
Automatic extraction	Yes (Crawler)	Manual	Manual	No
Never-translate enforcement	Automatic	Manual	Manual	No
Validation after AI translation	Yes	No	No	No
Bulk JSON import/export	Yes	CSV only	CSV only	N/A
DeepL glossary sync	Native	Manual	Manual	No
Versioning	Yes	Limited	Limited	No

Many TMS platforms offer glossary features, but they are largely passive — reference databases that translators are expected to consult manually. better-i18n enforces the glossary actively, at every point where a translation decision is made. The difference in outcome is significant.

Frequently Asked Questions

How many terms can a glossary contain?

There is no hard limit. Glossaries with tens of thousands of terms are supported. For AI translation, the system automatically selects the relevant subset of approved terms for each translation job to keep prompts efficient.

Can different projects have different glossaries?

Yes. Each project can have its own glossary, and you can also define organization-wide glossaries that apply to all projects. Terms in project-level glossaries override organization-level terms when there is a conflict.

What is the difference between a glossary term and a never-translate rule?

A glossary term has a source word and one or more approved target translations. A never-translate rule is a special type of glossary entry where the approved "translation" is the source word itself — it should pass through unchanged in all languages. You can also apply never-translate selectively to specific languages while providing translations for others.

Can the Context Crawler build my glossary automatically?

The Context Crawler proposes candidate glossary terms based on analysis of your product website (via Firecrawl) and your GitHub repository. Proposed terms enter the glossary as drafts — a human reviews and approves each entry before it is enforced. The crawler identifies repeated phrases, product names, and branded language, significantly reducing the manual work of glossary creation.

How does the DeepL sync work?

When you trigger a glossary sync, better-i18n converts your approved terms into DeepL's glossary format and uploads them via the DeepL API. DeepL then uses these terms in all subsequent translations for the configured language pairs. You can re-sync at any time after updating your glossary.

How do I handle terms that translate differently depending on context?

Each glossary entry supports context tags and domain labels. For terms with multiple valid translations depending on context (grammatical gender, register, domain), you can create multiple entries for the same source term with different context restrictions. The system selects the most appropriate entry based on the file or component context of the string being translated.

Can I import my existing glossary from another tool?

Yes. better-i18n supports glossary import from JSON format. Structure your terms with source text, target translations, and metadata, then use the bulk import feature to bring everything in at once. Imported terms start as drafts, giving your team a chance to review before enforcement begins.

Start Enforcing Brand Terminology Consistency Today

Glossary Management is available on all better-i18n plans. Start with the Context Crawler to automatically extract your first batch of candidate terms, review and approve them through the workflow, sync approved terms to DeepL, and watch them immediately start influencing your AI translation output.

For teams migrating from another tool, use the JSON import to bring your existing terminology database and start enforcing it immediately.

Start your free trial — or learn how Translation Memory complements glossary management to deliver complete consistency at the segment and term level.

Glossary Management: Consistent Brand Terminology Across Every Language