Translation & Localization
Phrase Localization Platform
EU-Made Report an error0–5 in half steps. 5 means the rubric's top anchor is met on the evidence.
by Phrase a.s. · phrase.com
Compare with Lokalise → Report an error on this page Is this your product? →
Link to this page
Add ?theme=dark to the image address for dark backgrounds.
The badge shows the category and the month of the latest evaluation, and EU-Made where
that is the product’s provenance. It never shows a score.
Phrase Localization Platform at a glance
- Vendor
- Phrase a.s.
- Headquarters
- Not stated on the vendor’s pages as of 1 Nov 2026
- Data hosting
- Not stated on the vendor’s pages as of 1 Nov 2026
- Free plan
- Not stated on the vendor’s pages as of 1 Nov 2026
- Panel verdict
- The AI panel rates Phrase Localization Platform 3.5/5, the mean of 6 judges. Phrase (Phrase a.s. how scores are made
- As of
- Vendor pages captured 1 Nov 2026; evaluated by the panel 2 Nov 2026. all sources
Phrase Localization Platform alternatives, European ones first →
Read this page as one judge. Each weighs the same scores by what they care about.
The panel's verdict
Phrase (Phrase a.s. of Prague, formerly Memsource) is a translation management and software localization platform that scores highest on machinery: machine translation, workflow and review, and integrations and formats each land at 7 to 8, on 30-plus machine translation engines with automatic selection, a CAT editor with in-context preview, real-time translation memory, nine versioned APIs and over 50 integrations. The weak side is what the pages leave unverifiable: pricing transparency is lowest at 5 to 6, since only the Team tier is public — 1,195 per month billed annually with unlimited TMS seats and 20 Strings seats — while Business and Enterprise read 'Custom' and the Studio add-on has no published price; content confidentiality and portability and exit follow at 5 to 7. That two-point spread is the closest the panel came to splitting: judges crediting ISO 27001, SOC 2 Type I and a complete subprocessor list scored top of the range; those wanting a written commitment that translated texts are not used to train models, or a documented import path, scored at the bottom — a difference in requirements, not evidence.
Speaks for it
- Workflow and review scores of 7 to 8, for a CAT editor with in-context preview, real-time translation memory, term bases and linguistic quality assessment profiles assignable per workflow step
- Integrations and formats scores of 7 to 8, for nine versioned APIs, CLI and client libraries, over-the-air updates with a cache-then-bundle fallback, and over 50 integrations including GitHub, Figma and WordPress
- Machine translation scores of 7 to 8, for 30-plus engines including DeepL and Amazon Translate, automatic engine selection, MT glossaries and a bring-your-own-engine route
- Translation memory exports as TMX or XLSX and term bases as TBX or XLSX through documented API endpoints, with customer content deleted within 60 days of permanent deletion or termination
- Security documentation covers ISO 27001 for Phrase a.s., SOC 2 Type I with the report on request, TLS 1.2 in transit, encryption at rest, and a subprocessor list naming the machine translation and AI providers
Held against it
- Pricing transparency is the weakest criterion at 5 to 6 — Business and Enterprise are 'Custom — Get in touch', the Phrase Studio add-on has no published price, and we found no public information on overage handling
- We found no public information on whether translated texts are used to train models, on zero-retention terms with the named LLM providers, or on restricting third-party AI processing per project
- The exit is export-only — we found no public information on importing translation memory or glossaries from another tool, on a migration path from named competitors, or on a self-hosted edition
- The stated architecture is public cloud and multi-tenant, the security page names AWS and Google Cloud data centers without a region, and we found no public information on a published customer data processing agreement
- Vendor-reported outcomes such as the 45% drop in post-editing and 40% less spend appear without a published translation-quality evaluation or its method
Best for
- You ship multilingual software and need GitHub and Bitbucket sync, key-based strings and over-the-air updates with a cache fallback so an outage cannot break the app
- You coordinate agencies and freelancers and need vendor management, cost tracking and per-step linguistic quality assessment in one platform
- You translate high volumes machine-translation-first and want 30-plus engines, automatic selection and glossaries governed by your own term bases
Avoid if
- You must compute your full annual cost before committing — only the Team tier is published, at 1,195 per month billed annually, while Business and Enterprise are 'Custom — Get in touch' and the Phrase Studio add-on is unpriced
- You require a private or dedicated deployment — the documented architecture is a public cloud, multi-tenant model
- You cannot have named third-party AI providers process your content — the subprocessor list names Google Ireland and Microsoft as machine translation and generative model providers, with locations and transfer mechanisms
- You need the hosting location confirmed before signing — the security page names AWS and Google Cloud data centers without a region, and the EU-hosted environment appears in the bug bounty scope only as an exclusion
The scores
| Criterion | Panel median |
|---|---|
| Machine translation & AI quality | 3.5 of 5 |
| Translation memory, terminology & review workflow | 4 of 5 |
| Integrations, file formats & developer tools | 4 of 5 |
| Confidentiality of translated content | 3 of 5 |
| Data export, portability & self-hosting | 3 of 5 |
| European sovereignty | 3 of 5 |
| Pricing transparency | 2.5 of 5 |
Machine translation & AI quality
Show reasoningHide reasoning
How this is scored
What the machine translation does and how well it is controlled: the languages and language pairs covered, the engine (own neural or LLM model, or a choice of third-party engines such as DeepL, Google, Microsoft or an LLM), glossaries and formality applied during translation, document translation that keeps the layout (Word, PowerPoint, PDF, InDesign), custom or adaptive models trained on the customer's own translations, automatic quality estimation or scoring, and published evaluations of translation quality.
0 — No machine translation of any kind, and no way to connect an engine.
3 — Machine translation is offered or one engine can be connected; languages, glossary use and quality controls are basic or not described.
5 — Machine translation for the major European and world languages with the list published, glossaries applied during translation, a formality or tone setting, document translation that keeps formatting for common office formats, and either an own engine or a choice of several named engines.
8 — Custom or adaptive models that learn from the customer's translation memory or corrections, automatic quality estimation per segment with a threshold that routes weak segments to human review, LLM-based translation with style guides and context, engine selection per language pair stated, and a published quality evaluation with its method.
10 — Translation quality that an organisation can measure and steer: evaluations against named benchmarks or blind human review published with their method, quality estimation calibrated and explained, custom models with their training data and isolation documented, and every machine-translated segment traceable to the engine, model version and glossary that produced it.
The Localization Manager
Thirty-plus engines with automatic selection plus bring-your-own, an adaptive engine that learns from our translation memories and term bases, glossary support, Auto Adapt and quality estimation — this is exactly the engine orchestration I need for fourteen languages, in 500+ languages. What holds it back is that I found no published quality evaluation with its method; the 45% post-editing drop and 40% less spend are reported outcomes, and per-segment quality estimation is described without the routing rules being spelled out. 1 4 6 12 14 17
The App Developer
The engine side is the strongest story here: orchestration that picks from a broad range of engines and LLMs or routes through my own, 30-plus MT engines including DeepL, Amazon Translate, ModernMT and NextMT, customisation of NextMT from my translation memories via Phrase Custom AI, MT glossaries with automatic engine selection, and quality estimation branded as Phrase QPS. It stays below the top marks because the quality numbers like the 45% drop in post-editing are vendor-reported outcomes, and we found no public information on a published translation-quality evaluation with its method or on quality-estimation routing of weak segments to human review. 1 4 12 14
The In-House Document Translator
The editor integrates with more than thirty named engines including DeepL, Google, Microsoft, Amazon Translate and the vendor's own NextMT, with automatic engine selection, glossaries applied during translation, and an adaptive engine that learns our terminology and tone from translation memories and term bases; custom models and quality estimation appear as plan features. Legal and HR document translation is offered with "support for every file format". What I still miss before sending a draft contract: we found no public information confirming that Word and PDF layout is preserved, and no published evaluation of translation quality with its method. 12 4 14 1 13 7
The Data Protection Officer
Adaptive and custom engine models that learn from the customer's own translation memories, quality estimation, glossaries, automatic engine selection across more than thirty named engines and five hundred-plus languages are all documented, with a bring-your-own route for our own engine. What I could not find is a published translation-quality evaluation with its method, so output quality cannot be verified from the public pages alone. 1 4 6 7 12 14
The E-Commerce Marketer
Automatic engine selection across 30-plus engines, a bring-your-own-engine option, DeepL and LLM routes, MT glossaries and unlimited languages on every plan are all documented, and the adaptive engine learns from my translation memory and term bases with quality estimation and Markdown style guides applied. For a shop with no translator on staff, that covers the machine doing the heavy lifting while I keep control of tone. We found no published quality evaluation with its method, so the reported 45 percent drop in post-editing stands as a headline figure and I stop just short of the top marks. 1 4 6 7 12 14 17
The Skeptic
The engine bench is real and named: 30-plus engines including DeepL, Amazon Translate and ModernMT alongside the own NextMT, with glossaries, automatic engine selection, quality estimation (QPS), style guides and LLM orchestration documented, and Custom AI adapting NextMT to the customer's translation memories. What I cannot find is a published quality evaluation with its method — the reported 45% post-editing and 40% spend figures appear without one — nor a stated mapping of engine to language pair or any per-segment traceability, so I stop one band short of the top. 1 4 6 12 14 17
Translation memory, terminology & review workflow
Show reasoningHide reasoning
How this is scored
How translation is managed as an ongoing process: translation memory with fuzzy matches, a termbase or glossary managed centrally, style guides, review and approval stages, automated QA checks (placeholders, tags, length, terminology, punctuation), an in-context or visual editor, screenshots and context for translators, tasks and deadlines, roles per language, and ordering or inviting human translators and agencies inside the product.
0 — No management of translations at all; text goes in and comes out with nothing stored or reviewed.
3 — Translations can be edited or a glossary kept; translation memory, review stages and QA checks are basic or not described.
5 — A translation memory reused across projects, a central glossary, at least one review or approval stage, automated QA checks for placeholders and terminology, comments and screenshots for context, and roles that limit translators to their languages.
8 — Configurable multi-step workflows with automation rules, an in-context or visual editor for apps or websites, QA checks configurable per project with blocking rules, style guides applied by the AI, translation memory and termbase shared across projects with penalties and priorities, vendor or agency management with ordering inside the product, and progress and cost reporting per language.
10 — A translation process an enterprise runs on: workflows per content type with conditional routing on quality scores, linguistic quality assessment with a named framework (MQM or similar), termbase with approval workflow and forbidden terms, full revision history per segment, and productivity and quality metrics per translator and vendor.
The Localization Manager
Real-time translation memory, centralized term bases, a CAT editor with in-context preview, multi-step workflows with an LQA profile assignable per step, vendor sharing on projects, cost tracking, Analytics included in plans and a codeless orchestrator — that is how I actually run two agencies and a freelancer pool. The captured pages stop short of naming the quality assessment framework, describing configurable blocking QA rules for placeholders, or stating memory penalties and priorities, so I hold it just under the highest mark. 7 13 14 15 16 19
The App Developer
This is the process control I swore I needed after hand-merging string files: real-time translation memory with auto-propagation, term bases carrying approved and forbidden terms, an LQA profile assignable to each workflow step, vendor sharing and vendor management, cost tracking with analytics, and a CAT editor with in-context preview and visual context for translators. It reads as a full enterprise-grade process on the captured pages, though we found no public information on a named linguistic-quality framework such as MQM, per-segment revision history, or productivity metrics per translator and vendor. 7 13 14 15 16 19
The In-House Document Translator
Translation memory works in real time, term bases carry approved and forbidden terms with domain and status filters, every workflow step can carry a linguistic quality assessment profile, and translators work in a CAT editor with contextual visualization and in-context preview. Vendor management, cost tracking, a codeless workflow orchestrator, versioned Markdown style guides and analytics are all documented. We found no public information on a named linguistic quality framework, per-segment revision history, or productivity metrics per translator. 14 20 15 7 2 16 17
The Data Protection Officer
Multi-step workflows with a linguistic quality assessment profile assignable per step, codeless workflow automation, an in-context preview, real-time shared translation memory and centralized term bases including a forbidden-term flag, vendor management and cost tracking are documented. We found no named quality assessment framework and no documented revision history per segment, which keeps a regulated buyer below the top anchor. 2 7 13 14 15 16 19 20
The E-Commerce Marketer
The CAT editor with in-context preview, real-time translation memory and central term bases with approved and forbidden term statuses would let me fix a bad product name by hand and keep it fixed, and multi-step workflows carry a linguistic quality assessment profile per step with a completion rule that blocks unfinished segments. Vendor management, Language Quality Assessment and customizable workflows already sit on the Team plan, with cost tracking and advanced reporting alongside. We found no public information naming an external linguistic quality framework or per-translator metrics, which is the only thing holding it back. 2 7 13 14 15 16 19 20
The Skeptic
Multi-step workflows with an LQA profile attachable to each step, a CAT editor with in-context preview and contextual visualizations, real-time translation memory, term bases, spellcheckers, vendor management, and cost tracking with a dedicated analytics product are all documented, alongside fine-grained linguist permissions per project. I found no public information on how automated QA checks are configured or on a named linguistic quality framework, so the review machinery falls short of the enterprise band. 2 7 13 14 15 16 19
Integrations, file formats & developer tools
Show reasoningHide reasoning
How this is scored
How content gets in and out: the file formats supported (XLIFF, PO, JSON, YAML, Android XML, iOS strings and xcstrings, Office documents, InDesign), a REST API and CLI, SDKs and over-the-air updates for apps, Git repository sync (GitHub, GitLab, Bitbucket) and CI pipelines, connectors for CMS, shop and design tools (WordPress, Contentful, Shopify, Figma), browser and office add-ins, webhooks, and a JavaScript snippet or proxy for website translation.
0 — No documented way to get content in or out except typing or pasting by hand.
3 — File upload and download or a basic API; formats, repository sync and connectors are basic or not described.
5 — A documented REST API, a list of supported file formats covering the common software and document formats, sync with at least one Git host or CI, and connectors or plugins for several named CMS, shop or design tools.
8 — A CLI and SDKs for named platforms, over-the-air updates for mobile or web apps, Git sync with branching, webhooks, connectors for the main CMS, shop, design and support tools, office or browser add-ins, a public status page, and API limits documented.
10 — Integration that scales across an organisation: an API covering every function of the product with versioning and SDKs, an app or connector framework third parties build on, continuous localization in CI with pull requests opened automatically, and every connector documented with its limits and the fields it syncs.
The Localization Manager
Over fifty integrations and fifty-plus file formats, GitHub with branches, Bitbucket, Figma, Contentful, Zendesk, a WordPress plugin that auto-creates projects and assigns translators or vendors, webhooks, CLI and client libraries, over-the-air updates with a cache-then-bundle fallback, and nine documented APIs with version pinning and rate-limit behaviour. I found no public information on office or browser add-ins, shop connectors or a status page, which is what separates this from a fully complete integration story. 2 3 5 6 14 16 17
The App Developer
As the developer on this panel, this is close to my wish list: CLI and client libraries, over-the-air updates for mobile apps with a cache-then-bundle fallback so an outage cannot break the app, GitHub and Bitbucket sync that pushes branches and returns finished translations to the source, webhooks, nine versioned APIs under a shared authorization model, MCP support, and 50-plus file formats and integrations covering Contentful, Figma, Zendesk and WordPress. It stops short of the top level because we found no public information on office or browser add-ins, a public status page, numeric API rate limits, or pull requests being opened automatically as part of the repository sync. 2 3 5 6 16
The In-House Document Translator
Nine documented APIs with version pinning and a changelog, a CLI with client libraries, over-the-air updates with a cache fallback, GitHub and Bitbucket sync, and connectors for Contentful, Figma, Zendesk, Drupal, Adobe Experience Manager, WordPress and Google Drive, SharePoint or Dropbox — with over fifty file formats and more than fifty integrations counted. We found no public information on a browser or office add-in or a public status page; the documented limit is the 429 response when Language AI's rate is exceeded. 6 5 16 17 2 3
The Data Protection Officer
Nine APIs with version pinning, a CLI with client libraries, webhooks, over-the-air delivery with a cache fallback, and more than fifty integrations covering GitHub, Bitbucket, Figma, Contentful, Zendesk and WordPress cover the connectors we would need, with a third-party-built category suggesting an ecosystem. We found no public information on office add-ins or a public status page, and rate limits are documented for one API only. 3 5 6 16 17
The E-Commerce Marketer
Over 50 integrations spanning CMS, repositories, design and support tools, nine documented APIs with version pinning and a changelog, CLI and libraries, over-the-air updates with a cache fallback, GitHub and Bitbucket branching, and a WordPress plugin with WPML support and a Yoast SEO tie-in that speaks directly to my multilingual search traffic. We found no public information on a public status page or exact API limits, and among all the named connectors we found no mention of Shopify, which is the one I need first. 2 3 5 6 16 17
The Skeptic
Nine documented APIs with pinnable versions, a CLI and client libraries, webhooks, over-the-air updates with a cache-then-bundle fallback, custom connector building, and 50-plus integrations covering GitHub, Figma, Contentful, Zendesk and WordPress — this is a genuine developer surface. We found no public information on a public status page, office or browser add-ins, or the actual API rate-limit numbers, and the 50-plus file formats are counted but not enumerated on the captured pages. 2 3 5 6 16 17
Confidentiality of translated content
Show reasoningHide reasoning
How this is scored
What happens to the text a customer translates: whether source and target texts are stored and for how long, deletion after translation, whether customer content is used to train the vendor's or a third party's models, which machine translation and LLM providers a TMS sends text to and whether that can be restricted or switched off, encryption in transit and at rest, single sign-on, roles and audit logs, and certifications (ISO 27001, SOC 2) with their scope. Where the text is processed is scored under sovereignty.
0 — Customer texts are stored indefinitely and used for the vendor's own purposes, or no statement on data handling exists at all.
3 — The product is described as secure or GDPR-compliant and a privacy policy exists; retention of translated texts, model training and third-party engines are basic or not described.
5 — A documented statement that customer texts are not used to train models, retention or deletion of translated texts stated, the third-party engines or LLM providers named, encryption in transit and at rest, single sign-on on a business tier, and a data processing agreement offered.
8 — Texts deleted immediately after translation or after a retention period the customer sets, third-party AI processing switchable per project or account, zero-retention terms with the named LLM providers, audit logs, granular roles, ISO 27001 or SOC 2 certification stated with its scope, and a private or dedicated deployment option.
10 — Confidentiality that can be verified: an on-premise or customer-controlled deployment, customer-managed encryption keys, every subprocessor that sees content listed with its role, independent audit reports available, and every data flow of a translation request documented end to end.
The Localization Manager
ISO 27001 for the Czech entity and SOC 2 Type I, TLS 1.2 in transit and encryption at rest, deletion of customer content within 60 days of permanent deletion or termination, machine translation and LLM providers named in a subprocessor list, audit logs on request and single sign-on from the Business plan. I found no public statement that customer texts are not used to train models, no way to switch third-party AI off per project, and no private deployment — the security page describes a public-cloud, multi-tenant architecture. 7 9 10 11
The App Developer
The security documentation is more concrete than most: TLS 1.2 in transit, LUKS or AES-256 at rest, ISO 27001 certification for the Czech entity, SOC 2 Type I with the report available on request, single sign-on from the Business tier, a named subprocessor list, and customer content deleted within 60 days of permanent deletion or termination. For confidential product strings I still want more steering, and we found no public information on a statement that customer texts are not used to train models, on restricting third-party AI processing per project or account, on zero-retention terms with the LLM providers, or on a dedicated or private deployment — the architecture is described as multi-tenant public cloud. 7 9 10 11
The In-House Document Translator
Deletion is stated — within 60 days after we permanently delete content or terminate the agreement — the machine translation and LLM providers that could see text are named with locations and safeguards, encryption in transit and at rest is documented, and single sign-on sits on the Business plan. But the promise I need in writing, that translated texts are never used to train anyone's models, is something we found no public information on, and the architecture is described as multi-tenant public cloud with audit logs provided by the vendor's engineers only on request. 9 11 10 7
The Data Protection Officer
The captured pages document TLS 1.2 in transit, encryption at rest, deletion of customer content within 60 days of permanent deletion or termination, single sign-on from the Business plan, audit logs available on request, and ISO 27001 for the named Czech entity plus SOC 2 Type I. As data protection officer I must flag that we found no public statement on whether translated texts are used to train models, no zero-retention terms with the named LLM providers, and the stated architecture is public-cloud multi-tenant with no dedicated deployment option documented. 7 9 10 11 19
The E-Commerce Marketer
ISO 27001 for Phrase a.s., SOC 2 Type I, TLS 1.2 in transit, disk-level encryption at rest, an audit log available on request, single sign-on from the Business tier, and deletion of content within 60 days of permanent deletion or termination are all stated, and the subprocessor list names exactly which machine translation and AI providers can see text. But we found no public statement that customer texts are never used to train the vendor's or a partner's models, and no per-project switch or customer-set retention, so my product names and margins rest on trust rather than a documented rule. 7 9 10 11
The Skeptic
Encryption in transit and at rest, an ISO 27001 certificate for the named entity, SOC 2 Type I with Type II expected 2026, single sign-on on the Business tier, and deletion of content within 60 days after permanent deletion or termination are stated, and the subprocessor list names Google and Microsoft as the machine translation and LLM providers. But the pages are silent on whether customer texts are used to train models and on zero-retention terms with those providers, the stated architecture is a multi-tenant public cloud with no dedicated option documented, and audit logs are only available upon request. 7 9 10 11
Data export, portability & self-hosting
Show reasoningHide reasoning
How this is scored
Whether translation assets stay the customer's and can leave with them: export of translation memory and termbase in open formats (TMX, TBX, XLIFF, CSV), export of all projects and files, import from another TMS, an API that can read every asset, the return or deletion of data on termination, open-source code and a self-hosted or on-premise edition.
0 — Translations cannot be exported; they exist only in the vendor's interface or engine.
3 — Translated files can be downloaded; export of translation memory and glossary, import and termination terms are basic or not described.
5 — Translation memory exportable as TMX and glossary as TBX or CSV, all project files downloadable, import of translation memory and glossaries from another tool, and an API that reads translations and keys.
8 — Complete export of every asset (memory, termbase, files, comments, history) in open formats, a documented migration path from named competitors, the data returned or deleted on termination under stated terms, and a self-hosted or on-premise edition with its own documentation.
10 — The customer owns the assets end to end: open-source code under a named licence that runs fully self-hosted with feature parity stated, a complete export another tool can import without loss, and the deletion of every copy on termination documented.
The Localization Manager
Translation memory exports to TMX with rich segment fields and term bases to TBX, both through the API, linguists can download the memory as TMX, and content is deleted within 60 days of termination — the core assets can leave. I found no public information on importing a memory or glossary from another tool, a documented migration path from named competitors, or a self-hosted edition, so the exit is only partial. 6 9 19 20 21
The App Developer
My exit path is documented where it counts: translation memory exports as TMX or XLSX and term bases as TBX or XLSX through documented API endpoints, an API that reads projects, jobs, keys, memory and terminology, and deletion of remaining data within 60 days of termination. We found no public information on importing translation memory or glossaries from another tool, a documented migration path from named competitors, or a self-hosted or open-source edition I could run myself if the vendor changed course. 6 9 20 21
The In-House Document Translator
Translation memory exports as TMX or XLSX and term bases as TBX or XLSX through the API, with term fields including forbidden and preferred, and content is deleted within 60 days of termination. We found no public information on importing memory or glossaries from another tool, on exporting comments or history, or on a self-hosted edition. 21 20 19 9 6
The Data Protection Officer
Translation memory leaves as TMX and term bases as TBX through documented export endpoints with selectable fields, linguists can download the memory as TMX themselves, and deletion within 60 days of termination is stated in the privacy policy. We found no public information on importing memories and glossaries from another tool, a documented migration path from a named competitor, or a self-hosted edition. 6 9 19 20 21
The E-Commerce Marketer
Translation memory exports as TMX or XLSX through documented API endpoints, term bases as TBX or XLSX with forbidden and preferred term fields intact, and linguists can download the memory themselves; content is deleted within 60 days after termination. We found no public information on importing memory or glossaries from another tool, a migration path from named competitors, or a self-hosted edition, so my exit plan is export-what-exists rather than a documented move. 9 19 20 21
The Skeptic
Translation memory exports as TMX or XLSX and the termbase as TBX or XLSX through documented API endpoints with rich metadata fields, linguists can download the memory as TMX, the APIs read projects, jobs, memory and termbases, and data is deleted within 60 days of termination. We found no public information on importing memory or glossaries from another tool, on complete export of files, comments and history, or on any self-hosted edition — so the memory can leave, but the six-year-old-spreadsheet question is only half answered. 6 9 19 20 21
European sovereignty
panel opinion
Show reasoningHide reasoning
How this is scored
Who the contracting entity and the parent company are, where texts are translated, stored and processed, and who the subprocessors are (hosting, machine translation engines, LLM providers, CDN for over-the-air delivery, support and payment tools). Independently sourced by the sovereignty pipeline; translated texts often hold personal data and trade secrets, so the whole chain counts, including the engines a TMS calls. A self-hosted edition the customer runs on its own European servers counts as the customer's chain, whatever the vendor's seat.
0 — Non-EU vendor and contracting entity, hosting unstated, subprocessors unnamed, and no data processing agreement offered.
3 — A privacy policy and a data processing agreement exist, but hosting location is unstated or non-EU and the subprocessors are not listed.
5 — EU hosting available and an EU contracting entity or EU representative, but parts of the chain (machine translation or LLM providers, CDN, support, payment) run through non-EU providers without a stated safeguard.
8 — EU hosting on named infrastructure, EU contracting entity, subprocessors listed with their locations including the engines and LLM providers, a data processing agreement published, and security certifications (ISO 27001, SOC 2) stated — or a self-hosted edition documented that keeps every text on the customer's own servers.
10 — Sovereign end to end and evidenced: European ownership, entity, hosting, engine and every subprocessor, or a self-hosted open-source edition with no call home, data export and deletion documented, and certifications published with their scope rather than implied.
The Localization Manager
The imprint names Phrase a.s. in Prague with a Hamburg GmbH, the subprocessor list gives locations and transfer mechanisms for Google, Microsoft, Rozetta and the other AI providers, and ISO 27001 and SOC 2 are stated. Hosting is on AWS with the region unstated on the captured pages — the EU environment only surfaces in the bug bounty scope — and a customer-facing data processing agreement is not evidenced, so I cannot confirm EU residency end to end. 8 10 11 18
The App Developer
The chain is documented to a level I can actually audit: a Czech contracting entity on Václavské náměstí with a commercial register file number, a full subprocessor list with addresses, locations and transfer mechanisms that names the hosting, machine translation and LLM providers including Google, Microsoft and Eleven Labs, and an EU-hosted environment visible in the bug-bounty scope. It falls short of the top level because hosting is described only as AWS and Google Cloud data centers without a named region, and we found no public information on a published data processing agreement. 8 10 11 18
The In-House Document Translator
The contracting entity is Phrase a.s. in Prague, formerly Memsource, with a German Phrase GmbH alongside, and the subprocessor list is unusually complete: Google Ireland and Microsoft named as machine translation and generative model providers, each with location and transfer mechanism. We found no public information on where the AWS data centres hosting the platform are located, an EU-hosted environment at eu.phrase.com appears only as an exclusion in the bug bounty scope, and no published data processing agreement was found — so I cannot say from these pages where a contract would actually be processed. ISO 27001 for Phrase a.s. and SOC 2 Type I are stated. 8 11 10 18
The Data Protection Officer
The imprint names Phrase a.s. in Prague and Phrase GmbH in Hamburg, the subprocessor list names the machine translation and LLM providers with locations and transfer safeguards (Google Ireland under a data processing agreement, Microsoft under standard contractual clauses, Rozetta in Japan under an adequacy decision), and ISO 27001 plus SOC 2 are stated. The security page names AWS and Google Cloud without naming a region for the main platform — the EU environment surfaces only as an exclusion in the bug bounty scope — and a published data processing agreement for customers is not confirmed on the captured pages. 8 9 10 11 18
The E-Commerce Marketer
The contracting entity is Phrase a.s. in Prague with a Phrase GmbH in Hamburg, and a public subprocessor list names hosting, machine translation and AI providers with their locations and transfer safeguards — as a Munich shop I can actually see the chain. But the captured pages name AWS and Google Cloud without naming a hosting region for where my texts sit, we found no published data processing agreement, and the bug bounty pages reveal an EU-hosted environment while keeping it out of testing scope. 8 10 11 18
The Skeptic
Phrase a.s. sits in Prague with a German GmbH beside it, and the subprocessor list is genuinely good — named providers with locations and transfer mechanisms, including Google Ireland and Microsoft EU/USA for machine translation and LLM work, Eleven Labs and Rozetta for voice and Japanese engines. But the security page names AWS and Google Cloud without naming a region, an EU-hosted environment surfaces only in passing on the bug bounty scope page, and we found no published data processing agreement on the captured pages. 8 10 11 18
Pricing transparency
Show reasoningHide reasoning
How this is scored
Whether a buyer can compute the real annual cost from public pages alone: price per tier by characters, words, keys, hosted words or seats, what happens when the volume exceeds the tier, the number of users, projects and languages included, which features (machine translation volume, AI credits, SSO, integrations, API access) cost extra, the free tier and its limits, the self-hosted edition and its support, minimum term and setup.
0 — No public prices at all; every customer is a sales call.
3 — A starting price or a free tier exists, but how cost scales with volume, seats or features is unclear.
5 — Tiers are public with their price per volume band and main inclusions, but at least one commonly needed piece (overage, extra seats, machine translation or AI volume, SSO, connectors) is unpriced or "on request".
8 — Every tier priced publicly with its scaling rule (characters, words, keys, seats), overage handling stated, add-ons and AI credits priced, the self-hosted edition's cost or licence stated, minimum term and VAT treatment stated.
10 — Complete price computability: annual cost derivable for a given volume, seat count and feature set, with every add-on, setup, support and overage fee and the cancellation and data-return terms published.
The Localization Manager
The Team plan is public at 1,195 per month billed annually with its volume bands spelled out — processed words, managed words, machine translation units and AI units per year, seats, portal users, unlimited languages and projects — plus a 14-day trial and 24/7 support. Business and Enterprise are 'Custom / Get in touch', the Studio add-on carries no price, and overage handling and VAT treatment are not stated, so a full annual cost is computable only for Team. 7
The App Developer
The Team plan is public at 1,195 per month billed annually with unusually granular published capacities — unlimited TMS seats and languages, 20 Strings seats, 2,500,000 TMS processed words, 1,200,000 managed Strings words, 12,000,000 machine translation units and 25,000 AI units per year — plus a 14-day trial and 24/7 support on all plans. But the Business and Enterprise tiers are 'Custom — Get in touch', the Phrase Studio add-on carries no published price, and we found no public information on overage handling or VAT treatment, so I can compute an annual cost only for the entry tier. 7
The In-House Document Translator
The Team plan is public at "1,195 per month billed annually" with its bands spelled out — 2,500,000 processed TMS words, 12,000,000 machine translation units and 25,000 AI units per year, unlimited TMS seats, 20 Strings seats — plus a 14-day trial and 24/7 support on all plans. The Business and Enterprise plans are priced "Custom" with "Get in touch", and we found no public information on what happens when the volumes are exceeded or what the Phrase Studio add-on costs. 7 3
The Data Protection Officer
The Team plan is published at 1,195 per month billed annually with unusually detailed annual volumes — 2,500,000 processed words, 12,000,000 machine translation units and 25,000 AI units — but Business and Enterprise read "Custom" with "Get in touch", the Studio add-on is listed on all plans without a price, and overage handling is not described. A 14-day free trial and the currencies offered are stated. 7
The E-Commerce Marketer
The Team tier is published at 1,195 per month billed annually with unusually concrete bands — 2,500,000 processed TMS words, 12,000,000 machine translation units, 1,200,000 Strings managed words per year, 20 Strings seats, unlimited TMS seats and languages, a 14-day trial and prices in four currencies, which would comfortably hold my catalogue without doubling when it grows. But Business and Enterprise are 'Get in touch', the Phrase Studio add-on carries no price, and we found no public information on overage handling, so I cannot compute a full annual cost on my own. 7
The Skeptic
The Team tier is public at 1,195 per month billed annually with unusually detailed capacity bands — Strings seats, managed words, TMS processed words, machine translation units and AI units — plus a 14-day trial and 24/7 support across plans. But Business and Enterprise are Custom with Get in touch, the Studio add-on carries no public price, and we found no public information on overage handling, VAT treatment or the Freelancer plan's price, so an annual cost is computable only for the Team tier. 7
European sovereignty — proven facts
0 of 6 dimensions provenBuilt only from facts shown on the vendor's own pages. A dimension we could not prove is left open, not scored as zero.
| Legal entity | Not determined ⚠ unverified | — | uncited Report an error |
|---|---|---|---|
| Hosting location | Not determined | — | uncited Report an error |
| Hosting options | Not determined ⚠ unverified | — | uncited Report an error |
| Open source | Not determined | — | uncited Report an error |
| Data processing agreement | Not determined | — | uncited Report an error |
| Transfer impact assessment | Not determined | — | uncited Report an error |
Questions about Phrase Localization Platform
Is Phrase Localization Platform European?
Yes. whats-best.ai classifies Phrase Localization Platform as EU-Made.
Where does Phrase Localization Platform store data?
Not stated on the vendor’s pages as of 1 Nov 2026.
Does Phrase Localization Platform have a free plan?
Not stated on the vendor’s pages as of 1 Nov 2026.
What are the best alternatives to Phrase Localization Platform?
As of 2 Nov 2026, the panel rates SimpleLocalize (3.5/5), LILT (3.5/5) and Tolgee (3.5/5) highest among the other 12 Translation & Localization products evaluated. Ratings that look equal are rounded to half stars; the order uses the unrounded rating, then the name. The highest-rated European alternative is SimpleLocalize (EU-Made). Ratings are the AI panel’s opinion, formed from public evidence: how scores are made. The full list, with European products first: Phrase Localization Platform alternatives.
Where this could be wrong
- Evidence ages. The oldest capture behind this page is from 1 Nov 2026. Vendors change pricing and policies without notice; every fact reflects its source as of the capture date shown in the registry.
- Weak sourcing — Legal entity, Hosting options. Not confirmed on the vendor’s own pages as captured.
- AI can misread a source. Extraction and judgement are automated; a citation guarantees traceability, not infallibility. If something here is wrong, say so — no account needed, every report is decided within 5 business days, and accepted corrections are published.
What we left out
A claim that does not survive our checks costs us the claim, not the page. This is what was taken off this one.
- 18 product facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 8 integrations facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 7 compliance facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 6 subprocessors facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 3 legal facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 1 hosting fact could not be confirmed on the vendor’s page as captured and was left out of this page and of the panel’s material. Know more? Tell us
- 2 sovereignty dimensions could not be confirmed on the vendor’s own pages and are shown as unknown. Know more? Tell us
Sources (21)
The pages every claim on this page was read from — each one checked, dated, and kept verifiable.
- 1 Vendor homepage phrase.com Checked 1 Nov 2026 Details →
- 2 Features page (translation management) phrase.com Checked 1 Nov 2026 Details →
- 3 Features page (software localization) phrase.com Checked 1 Nov 2026 Details →
- 4 Features page (machine translation) phrase.com Checked 1 Nov 2026 Details →
- 5 Integrations page phrase.com Checked 1 Nov 2026 Details →
- 6 API documentation developers.phrase.com Checked 1 Nov 2026 Details →
- 7 Pricing page phrase.com Checked 1 Nov 2026 Details →
- 8 Legal notice phrase.com Checked 1 Nov 2026 Details →
- 9 Privacy policy phrase.com Checked 1 Nov 2026 Details →
- 10 Security / trust page phrase.com Checked 1 Nov 2026 Details →
- 11 Subprocessor list phrase.com Checked 1 Nov 2026 Details →
- 12 Machine translation & AI quality — found from sitemap phrase.com Checked 1 Nov 2026 Details →
- 13 Machine translation & AI quality — found from sitemap phrase.com Checked 1 Nov 2026 Details →
- 14 Translation memory, terminology & review workflow — found from sitemap phrase.com Checked 1 Nov 2026 Details →
- 15 Translation memory, terminology & review workflow — found from sitemap developers.phrase.com Checked 1 Nov 2026 Details →
- 16 Integrations, file formats & developer tools — found from sitemap phrase.com Checked 1 Nov 2026 Details →
- 17 Integrations, file formats & developer tools — found from sitemap phrase.com Checked 1 Nov 2026 Details →
- 18 Confidentiality of translated content — found from sitemap phrase.com Checked 1 Nov 2026 Details →
- 19 Confidentiality of translated content — found from sitemap developers.phrase.com Checked 1 Nov 2026 Details →
- 20 Data export, portability & self-hosting — found from sitemap developers.phrase.com Checked 1 Nov 2026 Details →
- 21 Data export, portability & self-hosting — found from sitemap developers.phrase.com Checked 1 Nov 2026 Details →