# Is this translation good? Quality report with a traffic light per language

> Translate your business texts and see, language by language, what has been checked: intact placeholders, formatting, your glossary terms, register, length, figures and data, cultural risk, and back-translation. Each language comes with its green, amber, or red badge calculated with fixed rules, and each finding can be corrected directly from the report. Free and no registration to try it.

- **Source (source):** https://aitoolsforapps.com/google-translate-business-texts-glossary-275/f/translation-quality-report
- **Platform (platform):** AiToolsForApps
- **Language (language):** en

## In summary

- A Report That Tells You If You Can Send the Translation
- Each language comes with its badge —Verified, with warnings, or high risk— and behind it, a twelve-block report that shows you WHAT has been checked: placeholders, formatting, glossary, register, length, figures, cultural risk, and what the app has automatically corrected.
- Badge per language, before opening anything
- Deterministic verdict, not an opinion
- Twelve fixed blocks, always the same

## A Report That Tells You If You Can Send the Translation

The problem with machine translation isn’t that it’s bad; it’s that it looks equally good whether it’s correct or incorrect. A Korean text you can’t read seems correct by definition, and rereading it is useless. The only thing that helps is for someone to check specific things and tell you, one by one, what they’ve checked.

That’s what this report does. Each language in the translation run has its own report, always the same object with the same twelve blocks in the same order: the verdict with its one-sentence summary, and eleven fixed sections —placeholders and technical pieces, formatting, glossary, register, length, figures and data, «What will your customer read?», cultural risk, cultural adaptations, greetings and farewells, and automatic fixes—. Each section includes its count on the right: how many placeholders remain intact out of the original number, how many terms from your glossary were involved, how many characters and what percentage of expansion, how many warnings.

The badge is not a model’s impression; it’s calculated using declared deterministic rules. It turns RED if a numerical data point has changed value, been invented, changed currency, or had units converted, or if there’s a high severity cultural risk. It turns AMBER if there’s any other finding: an unguaranteed placeholder, mismatched formatting, mixed registers, an exceeded length limit, missing data, a cultural warning, or a meaning deviation in the back-translation. And it turns GREEN only if none of these issues are present. The same traffic light status is visible before opening anything, on the chip for each language, and on the result card.

Sections with findings come OPEN so you don’t have to search for them, while those in order come closed and green, but are still displayed positively: «the text contained no placeholders or technical pieces», «no terms from your glossary appeared in this text», «no known false friends or risky expressions in this language». The verified absence of problems is precisely what you’re looking for, which is why it’s stated, instead of leaving a silence that you can’t distinguish from an unchecked control.

And the report isn’t just a diagnosis; it’s where issues are fixed. Sections with a correctable finding include a button —correct the figure and keep the original, replace the risky segment with the proposed alternative, stop warning you about a specific word, revert to the literal greeting, launch the back-translation, retry the entire language— and applying any of these recalculates the verdict using the same rules. Below, you’ll find the traceability: the retries used (up to two, automatic, before you saw anything) and the list of interventions the app made on its own to present it this way.

## What's included

- **Badge per language, before opening anything** — Each language in the translation run displays its status —Verified, N warnings, or high risk— on the selector chip and the result card. A quick glance tells you where to start.
- **Deterministic verdict, not an opinion** — red if data changed value, was invented, changed currency, or had units converted, or if there’s a high cultural risk; amber for any other finding; green only if there are none. The rules are written and always the same.
- **Twelve fixed blocks, always the same** — verdict and summary, placeholders and technical pieces, formatting, glossary, tone of voice, length, facts and figures, what your customer will read, cultural risk, adaptations, greetings and closings, and automatic fixes.
- **What fails is open; what is good is in green** — sections with findings unfold automatically, and the rest remain closed with their count, so you don't read twelve sections to find the one that matters.
- **Green is written positively** — «no placeholders found», «no terms from your glossary appeared here», «no false friends in this language». A check performed and passed is stated; silence is indistinguishable from a control that no one has passed.
- **A one-sentence summary per language** — «Verified: placeholders, formatting, glossary, tone, and figures in order», or the exact list of what needs attention. This is what is exported and what you read if you only have ten seconds.
- **It's fixed directly from the report** — each correctable finding includes its action within its section, and upon applying it, the text updates and the verdict recalculates instantly. There's no need to re-translate to remove a warning.
- **Traceability and reuse** — you see how many automatic retries were used (up to two) and the list of what the app corrected on its own; the report travels in CSV and JSON exports, and the history can be filtered by quality status to find runs with warnings.

## Frequently Asked Questions

### How can I tell if a machine translation is good if I don't speak the language?

You can't know by reading it, and that’s the core problem: a text you don't understand always seems correct. What you can do is check things that don't depend on knowing the language. This tool performs a battery of such checks on each translation and shows them to you one by one: if your template placeholders are still there and not duplicated, if the number of headings, list items, links, code blocks, emojis, and paragraphs matches your original, if your glossary terms have been respected, if the formal or informal register is consistent throughout the text, if the character limit is exceeded, if every number, date, amount, and unit still holds the same value, if there are false friends or high-risk expressions in that specific language, and what the translation says when it's back-translated blindly into your language. What cannot be checked by code is not presented to you as checked.

### Does the AI decide the green badge, or are there rules?

There are rules, and they are declared. The verdict is calculated in code based on the findings from the eleven sections: it’s RED if a numerical data point has changed value, been invented, changed currency, or if units have been converted, or if there's a high cultural risk; it’s AMBER if there's any other finding (an unguaranteed placeholder, mismatched format, mixed registers, exceeded length limit, missing data, a cultural warning, or a deviation in meaning in the back-translation); and it’s GREEN only when there are none. The language model provides raw material for some sections—risk analysis, adaptations, back-translation—but it does not vote on the verdict. A green means «these specific checks have passed», not «this is a good advertising translation»: the style and legal responsibility for the text remain yours.

### What sections does the report include and what does each one tell me?

Twelve blocks, always the same and in the same order. Verdict and Summary: the badge and a sentence highlighting what needs attention. Placeholders and Technical Pieces: how many of your placeholders, tags, and links remain intact. Format: the inventory of the original versus the translation, heading by heading. Glossary: term by term, fulfilled, not guaranteed, or without a rule, with its explanation. Tone of Voice: whether the formal or informal tone is maintained throughout the text. Length: characters of the original, of the translation, expansion percentage, and limit. Facts and Figures: the complete list of data with its ✓ in green or its conflict in red. What Will Your Customer Read?: the literal back-translation and deviations in meaning. Cultural Risk: false friends and delicate expressions, with their severity. Cultural Adaptations: what has been intentionally changed and why. Greetings and Closings: the native formula used instead of the literal translation. Automatic Fixes: everything the app fixed on its own before giving it to you.

### If the report finds something, do I have to translate again?

No. Sections with a corrigible finding include a button within the report itself: revert a number to the original value, replace a risky fragment with the proposed alternative, silence warnings for a word that is normal in your sector, revert to the literal greeting instead of the native formula, launch the back-translation for that language, or retry the entire language if it failed. When applying a correction, the text is updated and the verdict is recalculated with the same rules, so you immediately see if the language is now green. And before all this, the app has already made up to two automatic retries on its own: what you see is the result after those corrections, and the ones it made are listed in the last section.

### Can I export the report from the app or consult it again?

Yes. The export includes a quality column and the number of warnings per language, and by checking the option to include the report, it adds a summary for each language and its automatic fixes: in CSV to open in a spreadsheet and in JSON with the full report if you want it within another system. Furthermore, each run is saved in the history with the quality status of each language, and the history can be filtered by that status to retrieve precisely the runs that had warnings or high risk and finish reviewing them.

- [Google Translate — Business Translations with Brand Glossary & Quality Report](https://aitoolsforapps.com/google-translate-business-texts-glossary-275)

## About the platform

**Google Translate — Business Translations with Brand Glossary & Quality Report** — an app by AiToolsForApps. Available at: https://aitoolsforapps.com/google-translate-business-texts-glossary-275/f/translation-quality-report

### Other languages

- `es`: https://aitoolsforapps.com/es/google-translate-textos-de-empresa-glosario-275/f/informe-de-calidad-de-la-traduccion
- `fr`: https://aitoolsforapps.com/fr/google-translate-textes-entreprise-glossaire-275/f/rapport-qualite-traduction
- `it`: https://aitoolsforapps.com/it/google-translate-testi-aziendali-glossario-275/f/rapporto-qualita-traduzione
- `pt`: https://aitoolsforapps.com/pt/google-translate-textos-empresariais-glossario-275/f/relatorio-de-qualidade-da-traducao
- `de`: https://aitoolsforapps.com/de/google-translate-geschaftstexte-glossar-qualitat-275/f/qualitaetsbericht-der-uebersetzung
- `bg`: https://aitoolsforapps.com/bg/google-translate-za-biznes-tekstove-rechnik-275/f/doklad-za-kachestvo-na-prevoda
- `ja`: https://aitoolsforapps.com/ja/google-translate-bijinesu-bunsho-yougo-shuu-275/f/hon-yaku-hinshitsu-repo-to
- `nl`: https://aitoolsforapps.com/nl/google-translate-zakelijke-teksten-termenlijst-275/f/kwaliteitsrapport-vertaling
- `ko`: https://aitoolsforapps.com/ko/google-translate-bijeuniseu-tekseuteu-yongeojip-275/f/beonyeok-pumjil-bogoseo
- `sv`: https://aitoolsforapps.com/sv/google-translate-foretagstexter-ordlista-275/f/kvalitetsrapport-for-oversattning
- `da`: https://aitoolsforapps.com/da/google-translate-virksomhedstekster-glosarlister-275/f/kvalitetsrapport-for-oversaettelse
- `ar`: https://aitoolsforapps.com/ar/google-translate-lnsus-alshrkat-alqamus-walmkhatba-waltqrir-275/f/taqrir-jawdat-al-tarjama
- `no`: https://aitoolsforapps.com/no/google-translate-for-bedriftstekster-ordliste-tiltaleform-og-rapport-275/f/kvalitetsrapport-for-oversettelse
- `fi`: https://aitoolsforapps.com/fi/google-translate-yritysteksteja-varten-sanasto-puhuttelu-ja-raportti-275/f/kaannoksen-laaturaportti
- `zh`: https://aitoolsforapps.com/zh/google-translate-business-glossary-275/f/fanyi-pinzhi-baogao
- `pl`: https://aitoolsforapps.com/pl/google-translate-do-tekstow-firmowych-glosariusz-forma-grzecznosciowa-i-raport-275/f/raport-jakosci-tlumaczenia
- `he`: https://aitoolsforapps.com/he/google-translate-ltkstim-askiim-milon-rmt-pniih-vdoch-275/f/doh-eikhut-tirgum
- `cs`: https://aitoolsforapps.com/cs/google-translate-pro-firemni-texty-glosar-osloveni-a-zprava-275/f/zprava-o-kvalite-prekladu
- `id`: https://aitoolsforapps.com/id/google-translate-untuk-teks-perusahaan-glosarium-kesopanan-laporan-275/f/laporan-kualitas-terjemahan
