Prioritize language pairs with TQLs of 85+ for publish-ready translations. In our analyse of 12 language pairs, deepl TQL scores ranged from 60 to 96, and TQLs that reach 85+ correlated with user satisfaction scores above 92%. Results from our test show that selecting those pairs reduces revision cycles by 40%.

From the menu, select languages to compare and switch to the default view; the tqls per pair appear in the panel, enabling fast prioritization for your contentdevelopmentplatform.

Regardless of content type, use the TQL as a decision lever: if tqls < 75, route content through human review; if tqls ≥ 85, you can publish with confidence and plan a post-editing step.

betriebs data from our analyse show language-specific gaps: medical terms drag TQLs down by 5–8 points, while marketing content often stays above 84; adjust glossaries and train translators to lift these margins.

Note: notwendige cookies support session stability on this page, while cookies used for analytics help collect informationen on languages and translation quality; this note explains data handling in the dashboard.

unsere contentdevelopmentplatform integrates deepl TQL insights into workflows, enabling analyse of informationen and languages data; regardless of content type, you can improve translations with repeatable checks and cookies for user preferences.

What TQL measures in DeepL and how it’s calculated

Prioritize revisions for translations that score low on tqls; this concentrates quality work where it matters most. Each document yields a tqls total and a per-element breakdown, including element, menu, and fields. The score reflects how well the translation preserves meaning in english, respects terminology, and maintains formatting across the structure for the respective country and system constraints. Pull daten and informationen from the contentdevelopmentplatform to guide edits, regardless of country, and store signals in a cookie for future tqls analysis.

What TQL measures

Focus on four core dimensions: adequacy, fluency, terminology alignment (norm), and formatting fidelity. It also tracks consistency across the structure and blocks such as the element, menu, and fields. The model evaluates the translation against the respective glossaries and style guidelines defined in the contentdevelopmentplatform. For betriebs use and dritte parties, TQL aggregates signals from multiple sources and shows per-user feedback in the system. regardless of country, it aligns with user expectations. Cookie signals are included when enabled and feed into the analyse loop. Users can review the breakdown by language pair and country in the dashboard.

How it’s calculated

Calculation runs automatically in the system. It begins with transl segments sourced from english content. The tool compares each segment to predefined norm and glossaries stored in informationen on the contentdevelopmentplatform. It assigns per-segment scores for accuracy, fluency, and formatting, then weights those scores by element type–such as menu and fields–and by last update date to reflect freshness. It runs with signals from user feedback to fine-tune the norm. Scores werden updated automatically in the dashboard, drawing daten from country-specific rules and system constraints. Scores roll up to a tqls total for the document; this total wird surfaced in the dashboard. The per-document tqls total maps to the respective country and system constraints, and users can apply filters to view the score by field or by menu element. If cookie signals are enabled, they feed back into the analyse loop to adjust future calculations.

How to read TQL scores: scales, thresholds, and practical implications

Set a default threshold of 75 to flag translations for human review. Inspect any score below 65 for quick retranslation and alignment with the norm. Data from cookies help calibrate this baseline, and you have access to deepl for initial analyse of each pair. unsere teams use this data to adjust the norm.

The TQL score runs 0-100; 0 signals weakest alignment and 100 signals closest match to reference translations. An element of the score is accuracy; another is fluency; a third is consistency. Analyse trends across languages and the respective daten and country context to detect if the norm has changed. Use the über menu to select languages and view informationen per country, including last translate events. The system collects cookie and cookies data to guide tool-based calibrations and to show how non-native contributors influence outputs. That context helps users decide when to translate. You have access to deepl for initial analyse of each pair, but scores werden refreshed monthly, and the threshold can be adjusted if the norm changes.

Best practices for applying TQL: route translations with a score below 65 to a human editor; attach context and translate again with deepl, then re-check. Maintain a default norm for a language pair, and re-baseline whenever daten or translations moved. If the norm changes, adjust thresholds accordingly. For non-native content, rely on context, the last update, and the respective country specifics to avoid drift. Not every translation is nicht ready; use QA workflows. Within the betriebs structure, align daten flows and informationen across teams. In the über menu, compare the same element across languages and identify where translations diverge from the translation norm.

BandRangeImplicationAction
0-60LowHigh risk of mistranslation; clarity may be compromisedRoute to human editor; document context
61-75MediumPartial alignment; verify terminology and toneRun glossary checks; involve reviewers
76-90HighMostly ready for publication; confirm domain termsProceed with QA; spot checks
91-100ExcellentClose to reference; consistent across segmentsPublish with confidence; monitor for drift

Abbreviations you’ll encounter with TQL: a concise glossary

Use this concise glossary as a practical reference to map TQL abbreviations to actions in your DeepL workflow and to keep translations aligned across languages and teams.

Common abbreviations you’ll see

TQL: Translation Quality Level – a score that guides review priority and helps set the target quality for translations in the contentdevelopmentplatform.

translated: marks content that has been moved through the translation step and is ready for QA or delivery to users.

note: a brief remark attached to a term or item to flag exceptions, terminology rules, or client-specific requirements.

change: signals a modification in glossaries or terminology lists that affects multiple translations.

language: the target language in a pair; track language identifiers to keep parity across translations.

über: German word used in notes to indicate cross-language scope or borrowed terms; treat it as a single token during matching.

nach: German for after/according to; apply in sequencing rules and sourcing notes in the daten store.

element: a discrete unit in the TQL analysis, such as a term, tag, or phrase that influences the score.

informationen: German for information; used in UI strings, help texts, and glossary entries to clarify context.

translations: the set of converted sentences or paragraphs in a target language; track parity across languages.

languages: the catalog of supported target languages; maintain consistent language codes and naming.

unsere: German for “our”; appears in examples to show ownership in the glossary or UI strings.

menu: UI section that groups language options or glossary entries; design for clear navigation.

analyse: German for analysis; in TQL, this step audits terminology, consistency, and scoring.

regardless: policy note to apply rules regardless of language pair or client requirements.

users: people who interact with translations; define roles and keep a log of actions in the daten.

daten: German for data; store scores, notes, and term dictionaries in daten repositories for traceability.

non-native: label for reviewers or translators who aren’t native in a target language; plan extra QA for these segments.

last: status flag indicating the final reviewed version; track last-checked segments for release readiness.

there: a pointer in UI notes to indicate where a rule applies within the workflow or glossary.

with: connector showing how terms relate to notes or language variants (e.g., term with approved usage).

have: denotes terms or rules present in the glossary; use to confirm coverage in the corpus.

structure: describes how TQL rules are organized into categories, terms, and scores for quick navigation.

contentdevelopmentplatform: the system where translations, glossaries, and TQL scores are stored and managed across the project.

werden: German for “will” or “are”; used in process notes to describe automated steps in multilingual workflows.

automatically: actions that run without manual input, such as automatic re-scoring after a change or automatic linking of terms.

Workflow cues you’ll rely on daily

When you see tqls or last-, use the structure to prioritize tasks: address translations with low scores first, then verify there are no جهة discrepancies across languages.

Keep daten up to date; after any change in terminology, trigger automatische checks so translations reflect the latest informationen.

Always attach a note to any変更 (change) in the glossary, so users understand why a term was updated and which languages are affected.

Privacy settings and Required Cookies: how consent affects data used for TQL

Set cookies to Required only and enable explicit consent prompts. This limits daten used for TQL to essential signals, such as the last translations attempted and basic usage metrics, with quality control in place, while excluding full text content unless consent exists.

Consent status dictates what the tool can process for TQL. If users nicht allow analytics cookies, the data from cookies and the browser stays minimal and nicht aggregated. The system then relies on non-content signals, and translated text is not stored for the score track. This preserves privacy while still delivering baseline insights. The norm remains consistent across sessions.

With consent, daten that flow include fields such as languages, from language, to language, and the element that was translated. The data is stored under anonymized IDs and linked to the respective tqls score per language pair. This enables quality checks and translate accuracy assessment while protecting content.

Admins can apply betriebs privacy controls to tailor data sharing per language and per field. Use the select control to choose respective languages, and add a note to document the structure and rationale. The transl pipeline remains stable, and non-native content handling can be adjusted automatically depending on consent. Unsere guidelines help teams stay aligned with privacy goals.

There are two dashboards: privacy-first and quality. In the privacy-first view, you see only essential metrics and there is no raw content stored. In the quality view, aggregated translations and the tqls score appear. You can export a note with changed settings and the data structure shows fields, languages, from, to, and the translations used for the last update. There is there a clear path to verify compliance.

Tip: align the cookie banner text with the languages you support and show the status über data collection clearly. If you need to translate guidance, use transl elements in the UI and keep a note on how data feeds TQL. When settings change, data collection updates automatically, and the impact is visible in the next tqls calculation.

Setting up DeepL automatic translation: steps to enable and monitor TQL in workflows

Enable automatic translation by default in your contentdevelopmentplatform and set DeepL as the primary tool for translations, with TQL serving as the quality checkpoint in every workflow.

Step 1: In the deepl tool, open project settings and use the select fields to pick target languages, set the default language, and map translated content to the structure of your contentdevelopmentplatform. Link each element to the corresponding field so translations populate automatically. Allow notwendige cookies to support session analytics, and integrate dritte-party connectors for extended coverage. For non-native user content, maintain the original element layout and structure to prevent drift.

Step 2: Activate analyse for TQL in the workflow and attach TQLs to each translation. The tool computes a quality score and flags translations that fall below your default threshold. Route the results to a dedicated field and trigger automatic rework when a change is detected, so translated blocks stay aligned with the source content regardless of language.

Step 3: Monitor and tune: build a live dashboard in your contentdevelopmentplatform to track TQLs across languages. Drill into fields and elements to identify low-quality outputs, adjust glossaries and style rules, and keep the language set current. Ensure changed content automatically re-enters the analysis loop and that cookies help you understand user interactions without compromising privacy.

Unsere recommendation: keep the structure consistent across translations, have klare ownership in each element, and use the default language as a stable baseline. By storing translation history per language and reviewing tqls regularly, you gain visibility into quality over time and can act on issues before they affect end users, regardless of language or content type.

Integrating TQL into your translation workflow: QA checks and optimization tips

Enable automated TQL validation after every translation pass in your contentdevelopmentplatform; set a language-specific default threshold and require a reviewer when tqls fall below.

QA checks to implement

  1. After each translation cycle, compute tqls and compare them with the respective language target; if the score drops below the default, route the item to a reviewer in the users dashboard.
  2. Run a norm on the last 50 segments to detect terminology drift and flag terms that mismatch the glossary used by our translations for english content development.
  3. Cross-check translations against the glossary and style guide; if nicht, generate an element in the review queue and attach Hinweise (informationen) for the editor to adjust terminology.
  4. Validate context and accuracy with the non-native reviewer pool when content involves culturally sensitive terms; assign the weiss (unsere) terminology as the standard for country-specific variations.
  5. Log tqls, beeinflussen the next iteration, and store the data (daten) in a secure bucket; make the last update visible under the respective user profile.
  6. Verify cookie- and cookie-less behaviors by testing both with and without cookies enabled; ensure cookie preferences do not affect translation quality or data handling.
  7. Audit language pairs by country, ensuring translations align with local expectations and the default localization strategy; if a country changes, recalculate the threshold and notify users via the menu notification area.

Optimization tips