OnlineDocTranslatorGuides

Answer

CAT Tool: What It Is, What It Does Inside a Translator's Workflow

A CAT tool is a desktop or cloud segment editor where translators work on bilingual segments with translation memory, term base, QA checks, and audit trail.

A CAT tool is the segment-editor layer a working translator sits inside all day — the screen that splits a source file into rows, pulls translation memory and machine translation into each one, and writes every keystroke to a bilingual audit trail. It is layer 2 of a seven-layer production stack, between the TMS that routes the job and the certifier who signs the exported file. This page defines what the tool is, what it does during production, and why the certified lane cannot be run without one.

Key points

  • A CAT tool is desktop or cloud software that segments a source document and lets a human translator edit each segment against translation memory, a term base, and inline QA checks — it never produces autonomous output the way an MT engine does.
  • CAT stands for Computer-Assisted Translation. The tool works segment by segment, keeping the human translator in control of every decision while the software surfaces matches, flags terminology, and enforces QA rules in real time.
  • At the center of every CAT tool is Translation Memory — a database that stores previously approved segment pairs and replays them automatically whenever the same or a similar source segment appears again. The first time a translator works on a new domain they will see most segments return as "New," but that proportion drops steadily as approved content accumulates in the TM over successive projects.
  • In a seven-layer localization stack, the CAT tool occupies Layer 2 — sitting above the TMS (Layer 1) that routes work and stores TM/TB data centrally, and below the MT engine (Layer 3) that can feed raw first-draft output back into the editor for post-editing.
  • memoQ and SDL Trados Studio are the two CAT tools most paired with a TMS on ISO 17100-aligned certified lanes, because both carry a segment-level audit trail and a reliable offline working mode — both requirements for legally defensible certified translation delivery.
  • The standard client-agreed fuzzy-match threshold for TM pre-translation is 75%: segments that match an existing TM entry at or above that score are pre-populated before the translator opens the file.
  • US per-word production pricing: translation $0.10–$0.25 per source word (or
    • A CAT tool is desktop or cloud software that segments a source document and lets a translator edit each segment against translation memory, a term base, and inline QA checks.
    • CAT stands for Computer-Assisted Translation — the tool assists a human translator segment by segment and does not translate autonomously the way an MT engine does.
    • memoQ and SDL Trados Studio are the CAT tools paired with a TMS on ISO 17100-aligned certified lanes, because both carry segment-level audit trail and offline working mode.
    • US per-word production grid: translation $0.10–$0.25 per source word, review $0.04–$0.08 per word, LQA $40–$90 per 1,000 words against an MQM-style scorecard.
    • The standard client-agreed fuzzy threshold for TM pre-translation inside the editor is 75% match.

    5–

    What is a CAT tool?

    A CAT tool is desktop or cloud software that splits a source document into segments and lets a translator edit each one against translation memory, term base hits, and inline QA checks, writing every change to a bilingual audit trail. It is one entry in a compared set covered in CAT Tools Compared: How Translation Memory Software Works, and it is distinct from both the TMS that routes the project and the MT engine that drafts first-pass output.

    What does a CAT tool actually do during translation?

    A CAT tool segments the source file, pre-populates each target segment from translation memory and machine translation, enforces the term base, runs inline QA checks on numbers, tags and placeholders, and exports a clean bilingual file back to the TMS. The seven steps a translator triggers inside the editor are:

    1. Open the project package handed down from the TMS.
    2. Let the tool segment the source by punctuation and tag rules.
    3. Accept TM leverage above the agreed fuzzy threshold.
    4. Post-edit the MT first draft segment by segment.
    5. Resolve term base flags as they appear in the target column.
    6. Clear every inline QA warning before locking the segment.
    7. Export the bilingual file and the clean target for review.
    The CAT tool workflow runs in 8 steps: the TMS delivers the source file; the CAT tool segments it; TM pre-fills matches at or above 75%; MT fills empty segments; the translator post-edits each segment; the term base and inline QA flags surface warnings; cleared segments are exported as a bilingual file; the review and LQA gate approves release. fills empty segments]
     D --> E[Translator post-edits each segment]
     E --> F[Term base + inline QA flags]
     F --> G{All warnings cleared?}
     G -- No --> E
     G -- Yes --> H[Bilingual export back to TMS]
     H --> I[Review and LQA gate]

    How does segmentation work inside the editor?

    The CAT tool breaks the source document into translation units — sentences — using segmentation rules keyed to punctuation, format tags, and language-specific exceptions, then displays each unit as an editable row. The segment-by-segment production surface is already described in detail on Localization Workflow: How Translators and Agencies Wire the Production Pipeline.

    How does translation memory leverage change the first-draft screen?

    Translation memory pre-fills segments that match stored past translations above the client-agreed fuzzy threshold — 75% — so the translator post-edits instead of typing from scratch. The CAT tool colour-codes exact, fuzzy, and no-match segments so the eye finds leverage first. The underlying mechanics are covered in How CAT Tools Use Translation Memory to Speed Up Your Work.

    How does the term base enforce approved terminology in real time?

    The term base holds client-approved renderings of product names, feature names, and regulated terminology. The CAT tool highlights source terms as the cursor enters each segment and flags deviations in the target as the translator types. The translator never alt-tabs to a glossary. Real-time term base flagging is covered Content Localization: How Translators and Agencies Produce It.

    What inline QA checks run while the translator works?

    The CAT tool runs inline QA on numbers, dates, placeholders, tag integrity, double spaces, untranslated segments, inconsistent translations, and term base violations. The warnings appear beside each segment and must clear before export. The nine checks a certified lane will not release without are:

    • Number mismatches between source and target.
    • Date format drift against the locale convention.
    • Placeholder count and family consistency.
    • Tag integrity (opening/closing, order).
    • Double spaces and stray whitespace.
    • Untranslated segments left in the source language.
    • Inconsistent translations of the same source segment.
    • Term base violations against approved renderings.
    • Punctuation mismatches at segment boundaries.

    What are the main features of a CAT tool?

    A CAT tool bundles seven core features: segmentation, translation memory, term base, inline QA checks, bilingual export, concordance search, and a segment-level audit trail — each one tied to a specific stage of the production workflow.

    FeatureWhat it doesWorkflow stage it serves
    SegmentationSplits the source into translation units by punctuation and tag rules.Project open
    Translation memoryPre-fills segments from stored past translations at or above the fuzzy threshold.First-draft screen
    Term baseFlags source terms and enforces approved target renderings in real time.Translation
    Inline QA checksBlocks tag, number, placeholder and terminology errors segment by segment.Translation and review
    Bilingual exportWrites a source+target file that review tools and auditors can read.Delivery
    Concordance searchLooks up prior translations of substrings across the TM for context.Translation
    Segment-level audit trailRecords who changed each segment, when, and from what prior value.Certified-lane sign-off

    Which CAT tool features matter for certified translation production?

    Four CAT features gate certified work: segment-level audit trail, offline working mode, named-translator attestation on export, and a locked bilingual review view. memoQ and SDL Trados carry all four, which is why they are the pair agencies run on ISO 17100-aligned lanes. The deeper Trados breakdown is in Trados as a CAT Tool: Features, Costs, and Honest Trade-offs.

    What are examples of CAT tools translators actually use?

    Working translators and agencies use five main CAT tools: SDL Trados Studio, memoQ, Phrase, OmegaT, and Wordfast — with memoQ and SDL Trados dominating certified and regulated lanes, and OmegaT holding the free open-source slot.

    ToolDeploymentLicenseCertified-lane fit
    SDL Trados StudioDesktop (Trados Team is cloud)PaidPrimary certified-lane choice
    memoQDesktop + cloud (memoQ cloud)PaidPrimary certified-lane choice
    PhraseCloud / browserPaidWeak for offline certified work
    OmegaTDesktopFree, open sourcePossible for low-stakes work
    WordfastDesktop + web (Wordfast Anywhere)Paid; Anywhere freeVaries by edition

    Which CAT tools are free or open source?

    OmegaT is the main free, open-source CAT tool used by professional translators. It carries TM, term base, and QA checks but lacks the vendor-support and ecosystem depth paid tools offer. The full capability-and-limit picture is in OmegaT for Professional Translators: What It Does and What It Lacks.

    Which CAT tools run in the browser (cloud / online)?

    Phrase and the cloud editions of memoQ and Trados Team run inside a browser, letting multi-translator projects share one TM in real time without local file transfer. The trade-off is weaker offline control for certified lanes. The day-to-day feature set is covered Phrase CAT Tool: Features Translators Use Every Day.

    Services

    Where does the CAT tool sit in the broader translation stack?

    The CAT tool is layer 2 of a seven-layer production stack: the TMS routes projects and stores TM; the CAT tool is the translator workbench; the MT engine produces first drafts; QA and LQA gate release; connectors move files; the certifier signs off. The broader program view is Localization Management: How Translators and Agencies Run Multi-Locale Programs.

    Why do certified translation lanes depend on the CAT tool's audit trail?

    Certified translation lanes depend on the CAT tool because the segment-level audit trail, bilingual review view, and offline file control produce the court-admissible chain of custody that USCIS filings, court exhibits, and registry submissions require. A cloud-only web editor cannot generate the same evidence. The four features a certifier checks before signing are:

    • Who edited which segment, with timestamps.
    • Prior-value history on every change.
    • A bilingual export locked against post-sign-off edits.
    • An offline working copy outside third-party cloud storage.

    What does USCIS-grade certified production cost per page when a CAT tool drives the lane?

    USCIS certified translation sits at $0.10–$0.25 per source word or $25–$150 per standard page of about 250 words, with the certificate of accuracy mandatory on every page and notarization an optional $10–$50 add-on. The CAT tool's exported bilingual file becomes the attachment of record.

    Production stageUS rateWhat it buys
    Translation$0.10–$0.25 / source wordHuman translator working inside the CAT editor.
    Review$0.04–$0.08 / source wordSecond linguist checks every segment.
    LQA$40–$90 / 1,000 wordsSampled MQM-style score against ISO 17100 criteria.
    Certificate of accuracyMandatory, per pageTranslator attestation attached to delivery.
    Notarization$10–$50 (optional)Notary witnesses the translator's signature.
    Standard page$25–$150About 250 source words.

    How it works

    What are the limits of a CAT tool — and where does the workflow break?

    1

    Custom-nested JSON that the stock filter cannot parse, causing segments to be skipped or merged incorrectly.

    2

    Segment drift when the source file is re-exported or updated mid-project, invalidating existing TM matches and forcing re-analysis.

    3

    Mixed placeholder families (ICU + sprintf + Angular, for example) in a single file, which generate QA warnings the translator cannot clear without developer intervention.

    4

    Hard-coded line breaks that split grammatical sentences into fragments, degrading both TM leverage and output fluency.

    5

    Locked tags whose integrity check blocks export, stalling delivery until the tag structure is corrected in the source or the filter rule is overridden with project-manager sign-off.

    How we use the CAT tool, and what we refuse to do with it

    We treat the CAT tool as the evidence-producing layer of the certified lane, not as a translation engine. Every certified job runs through memoQ or SDL Trados on an offline working copy, with the segment-level audit trail enabled from the first segment and retained with the delivered bilingual file.

    We do not sign a certificate of accuracy from a cloud-only web editor that cannot produce a locked bilingual export and a per-segment edit history. The attestation names a human translator, and the file beneath it has to be defensible if a court or an immigration officer asks who changed what and when.

    We refuse to pre-translate below the 75% fuzzy threshold and present the output as a human draft. Below that cutoff the TM match is a distraction, not leverage, and quoting it inflates the leverage column on an invoice. The post-editor starts from an empty target, and the review gate is priced at $0.04–$0.08 per source word on every segment.

    We will not run a certified lane on a free CAT tool because our clients need a vendor-supported filter chain and a documented QA profile. OmegaT remains fit for low-stakes work; it is not what we attach a certificate of accuracy to.

    A CAT tool splits a source document into segments and lets a human translator edit each one against translation memory, a term base, and inline QA checks, with every change written to a bilingual audit trail.

    Frequently asked questions

    What are CAT tools for translators?

    CAT tools for translators are software applications that support the translation process by breaking source documents into segments and presenting them alongside a workspace for writing target-language equivalents. They store every translated segment in a Translation Memory (TM), so repeated or similar text is auto-populated in future projects. Most tools also integrate a termbase (glossary), quality-assurance checks, and — increasingly — machine translation engines. Common examples include SDL Trados Studio, memoQ, Wordfast, Déjà Vu, and Smartcat.

    What does a cat tool do?

    A CAT tool splits a source document into sentence-level segments, displays them one at a time, and lets the translator write or edit the target text in a structured editor. It simultaneously queries a Translation Memory to surface matches from previously translated content — exact matches (100%) are reused as-is, while fuzzy matches (typically 75–99%) are presented as editable suggestions. Integrated termbases flag key terms, built-in QA checks catch omissions or inconsistencies, and the finished file is exported in the original format. The cumulative effect is faster throughput and higher consistency across a project.

    What is a CAT tool in translation?

    In translation, a CAT tool (Computer-Assisted Translation tool) is a professional productivity environment that combines a segmented editing interface, Translation Memory, terminology management, and QA automation in a single platform. It is distinct from machine translation: the human translator controls every decision, while the tool handles repetitive retrieval and format reconstruction. CAT tools are industry-standard in commercial, legal, technical, and certified translation workflows, and most translation agencies require deliverables to be produced in a supported CAT tool format (e.g., a Trados package or an XLIFF file).

    What is the phrase CAT tool?

    "CAT tool" is industry shorthand for Computer-Assisted Translation tool, a software category used by professional translators. The phrase distinguishes this class of productivity software — which keeps the human translator in control — from fully automated machine translation (MT) systems. In professional settings the phrase is used interchangeably with "translation environment tool" (TEnT) or "translation memory tool." You will encounter it in job postings, agency style guides, and ISO 17100 compliance discussions.

    What are CAT tools examples?

    The most widely used CAT tools in professional translation today are SDL Trados Studio, memoQ, Wordfast Pro, Déjà Vu X, Smartcat, and Memsource (now Phrase). Each supports Translation Memory and termbase management as core features. Smartcat and Phrase are cloud-native SaaS platforms, while Trados and memoQ offer both desktop and server/cloud editions. Specialised tools such as OmegaT (open-source) and Transifex (software localisation) also fall into the broader CAT tool category.

    Which is better, Trados or memoQ?

    Neither is objectively better — the right choice depends on workflow, budget, and client requirements. SDL Trados Studio has the largest market share and is the de-facto standard requested by large LSPs and enterprise clients, making it the safer choice if you want maximum compatibility. memoQ is widely preferred for its project-management features, more intuitive interface, and flexible licensing; it is especially popular in Europe. Both support XLIFF interchange, so files can be exchanged between them. Freelancers often hold licences for both tools.

    What is SMT and NMT?

    SMT (Statistical Machine Translation) and NMT (Neural Machine Translation) are two generations of automated translation technology that power the MT engines often embedded in CAT tools. SMT, dominant through the 2000s and early 2010s, used probabilistic models trained on parallel corpora to predict the most likely word sequences. NMT, which became mainstream around 2016–2017, uses deep neural networks (typically transformer architectures) to produce more fluent, context-aware output. Modern engines — including DeepL, Google Translate's neural mode, and ModernMT — are all NMT-based and significantly outperform legacy SMT systems on most language pairs.

    What are the two main types of translation?

    The two main types of translation are human translation and machine translation (MT). Human translation is performed entirely by a qualified linguist and is required for certified, legal, medical, and literary content where accuracy and accountability are non-negotiable. Machine translation uses algorithms (today predominantly neural models) to produce output automatically, as a first draft that a human post-editor then refines — a process called MTPE (machine translation post-editing). Within human translation, further distinctions are made between general, technical, sworn/certified, and literary translation.

    Where can I find CAT tools?

    CAT tools are available directly from their developers' websites: SDL/RWS for Trados Studio, memoQ for memoQ, Phrase for Memsource, and Smartcat.ai for Smartcat, among others. Most offer free trials of 30 days or a perpetually free tier (Smartcat). OmegaT is a fully free, open-source option downloadable at omegat.org. Professional translator communities such as ProZ.com and the SDL Community host comparison guides and discounted licences negotiated for freelancers. Some agencies provide licences to their regular contractors.

    Can I use Smartcat for free?

    Yes, Smartcat offers a permanently free plan for individual translators and small teams. The free tier includes unlimited Translation Memory storage, basic CAT editor access, and the ability to collaborate on projects. Paid plans add features such as advanced automation, MT integrations, and priority support, with pricing scaling by team size and monthly word volume. For freelancers doing occasional certified translation work, the free tier is often sufficient to manage standard projects.

    Related