memoQ

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.

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.

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.

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.

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.

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.

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.