>
xRM Translation Bridge
International Dynamics 365 organizations share one CRM across many languages: a sales team in Poland, service agents in Spain, account managers in Germany. The user interface can be localized — but the data cannot. Every description, note and free-text field stays in the language it was written in.
This insight explains what real-time data translation means in Dynamics 365, how it works under the hood, and what to look for before you choose a solution.
Real-time data translation in Dynamics 365 is the automatic translation of field values — not labels or metadata — at the moment a user opens a record, into that user's own language. Microsoft localizes the interface and translates labels, but field content such as account descriptions, case details or opportunity names is not translated natively. A data translation add-on detects the language of each text field, sends it to a translation provider such as Azure AI Translator, DeepL or Azure OpenAI, and shows the result next to the original. Well-designed solutions keep every translation as a Dataverse row, so repeated text is translated once, results are auditable, and customer data goes only to the AI resource the organization itself controls.
Labels & metadata
Table and field names, forms, views, option sets, sitemap entries. Translated once, deployed with the solution — the job of xRM Translation Studio.
“Account Name” → “Firmenname”Field data
Descriptions, notes, e-mails, case details, opportunity names. Written every day, in every language — translated when someone needs to read it. The job of xRM Translation Bridge.
“Presupuesto aprobado” → “Angebot genehmigt”When a record opens, the text fields are checked for their language. Fields already in the user's language, very short values, addresses, e-mails, URLs and phone numbers are skipped.
The remaining fields go to the translation provider in one batched request — one call per record instead of one per field keeps it fast. Rich text needs a provider that preserves formatting.
Results are stored so the same text is never paid for twice. Keyed on the source text rather than the record, one translation serves every record that contains it.
Users copy field content into public web translators and read the result in another tab.
Add-ons that translate configured columns on demand.
A Dynamics 365 add-on that translates field data the moment a record opens — with zero column configuration.
Translation quality, speed and cost differ widely between providers — and the differences show up exactly where CRM text lives: abbreviations, company names, payment terms. The numbers below come from our own test corpus of Polish, Spanish and French CRM records, measured in September 2026 through the full product path.
| Provider | Business languageabbreviations, company names | First translation10 fields, median | From cacheyour Environment | Cost per 1,000 recordslist prices |
|---|---|---|---|---|
| DeepLclassic machine translation | Goodresolves most abbreviations | 0.6 s | 0.2 s | ≈ €29 |
| Azure AI Translatorclassic machine translation | Mixedkeeps abbreviations as they are | 0.6 s | 0.1 s | ≈ $15 |
| OpenAIgpt-4o-mini | Very goodkeeps names, resolves abbreviations | 2.8 s | 0.1 s | ≈ $0.23 |
| Azure OpenAIyour Azure region | Very good | 3.2 s | 0.2 s | ≈ $3.85 |
| AnthropicClaude | Very good | 5.3 s | 0.2 s | ≈ $5.39 |
| GoogleGemini | Goodlarge language model * | 2–5 s * | 0.2 s | ≈ $0.80 |
Measured through the complete product path — plug-in, licence check, exclusions, language detection and cache included — with ten fields per record. Cached translations are served without any provider call, so a record that has been seen before costs nothing and appears in a fraction of a second. In a typical organisation roughly 70 % of translations come from the cache, which cuts the figures above accordingly. Cost is list price at the time of measurement and is billed by your provider, not by xRM Products. * Google was not part of this measurement run; the figures shown are its list price and the range we measured across comparable large language models.
Balanced — the default
The only provider in our test that combines strong business-language quality with a first translation well under a second. Formatting in rich text stayed intact in every case.
Best for notes and abbreviations
Large language models understood shorthand such as invoice, tax-number or contact-report abbreviations, and left company names, legal forms and addresses untouched — at the price of a slower first translation.
Lowest cost
Small models undercut classic machine translation by a wide margin, because language detection costs almost nothing — with DeepL, detecting a language costs a full translation.
Data residency
Both run in the region of the resource you create. DeepL processes in the EU. The other providers process in their own infrastructure — the settings page states this per provider.
You are never locked in: the provider is a setting, and every user can be given the right to switch for a single translation. Because translations are kept as Dataverse rows, switching provider does not mean paying for the same text twice.
Field content should go only to an AI resource your organization owns — your key, your region, your data processing agreement — never through the vendor's own service.
Column-by-column configuration breaks every time a table or field is added. Look for coverage of all tables by default, with exclusions instead of permission lists.
Translation quality differs by language pair and text type. Business shorthand and slang favour some providers, formatted rich text others — a choice protects quality and cost.
Without a cache, the same payment terms on three hundred records are translated three hundred times. Stored translations cut cost and make results auditable.
Exclusions should be enforced on the server and fail closed: if a check cannot be evaluated, the field is not sent.
Translation should start after the form has loaded and never block it. From a cache, a translated record should appear in well under a second.
| Browser translator | Per-entity add-on | xRM Translation Bridge | |
|---|---|---|---|
| Works without column setup | ✓ | — | ✓ |
| Stays inside Dynamics 365 | — | ✓ | ✓ |
| Automatic language detection | ✓ | — | ✓ |
| Choice of AI provider | — | — | ✓ |
| Translations kept & reusable | — | — | ✓ |
| Admin exclusions | — | ✓ | ✓ |
| Customer data only to your own AI resource | — | ✓ | ✓ |
xRM Translation Bridge turns multilingual data from a copy-paste chore into a one-click read — inside the record, in your language.
Welcome to the fascinating world of
xRM Products, get to
know us and our products now!
© 2026 xRM Products LLC
