Translation & Quality Evaluation Standards
ISO 17100, ISO 18587 and ISO 5060:2024 explained
These standards address different parts of professional translation services. ISO 17100 covers human translation and revision; ISO 18587 covers full human post-editing of machine-translation output; and ISO 5060:2024 provides guidance for evaluating translation quality.
Termwise uses controlled translation and evaluation processes to help clients choose the right service for their content: human translation and revision, full machine translation post-editing, or a structured assessment of existing translations.
What ISO 17100 covers
What ISO 18587 covers
What ISO 5060:2024 covers
The practical difference
How Termwise applies ISO-controlled workflows
Three standards, distinct roles in translation quality
ISO 17100 is the reference point for professional human translation services. It focuses on core translation processes, resources and the requirements needed to deliver a quality translation service.
ISO 18587 is the reference point for full human post-editing of machine-translation output. It is relevant when MT or AI-assisted output is used as the starting point and the final text must be reviewed and corrected by a qualified human post-editor.
ISO 5060:2024 provides guidance for evaluating translation output. It supports structured assessment of human translations, post-edited machine translations and unedited machine-translation output, using agreed criteria and quality ratings.
What ISO 17100 Covers
ISO 17100 provides requirements for the core processes, resources and other aspects necessary for delivering a quality translation service that meets applicable project specifications.
In practical terms, it is the standard most closely associated with professional human translation workflows: qualified linguists, clear project requirements, translation, revision and final delivery checks.
Qualified Human Translators
Translation work is performed by linguists with appropriate language competence, translation competence and subject-matter awareness.
Defined Project Requirements
The workflow starts from agreed specifications: language combination, audience, purpose, terminology, style and delivery requirements.
Translation and Revision
The translation is produced by a human translator and reviewed for accuracy, completeness, terminology and target-language quality.
Final Quality Checks
The final text is checked for formatting, names, numbers, dates, omissions, instructions and delivery-readiness.
What ISO 18587 Covers
ISO 18587 provides requirements for the process of full human post-editing of machine-translation output and for the competences of post-editors.
This standard is relevant when machine translation or AI-assisted translation output is used as the starting point. The human post-editor must review, correct and improve that output so the final text is suitable for the agreed purpose.
MT/AI Output as Starting Point
The workflow begins with machine-translation output, not a translation written from scratch by a human translator.
Qualified Human Post-Editing
A trained human linguist reviews and corrects the output using the source text, client instructions and intended use.
Full Post-Editing
The aim is not merely to make the text understandable, but to produce a usable, accurate and professionally controlled final text.
Suitability Assessment
Not all content is suitable for MT/AI-assisted workflows. Risk, subject matter, quality expectations and confidentiality need to be assessed first.
ISO 5060:2024 — Translation Quality Evaluation
ISO 5060:2024 provides guidance for evaluating human translations, post-edited machine translations and unedited machine-translation output. It addresses evaluator competence, sampling and an analytical approach that classifies errors and uses penalty points to produce an error score and quality rating.
Agreed Evaluation Scope
The intended use, assessment criteria and coverage are defined before evaluation, including whether the complete text or a selected sample will be assessed.
Qualified Human Evaluation
Linguists compare the translation with its source and assess identified errors against agreed criteria, with severity determined by their impact on the content’s intended use.
Documented Findings and Scoring
Findings provide a clear audit trail. Where commissioned, agreed scoring rules produce a quality score and acceptance rating.
How Termwise Applies It
Termwise evaluates translation quality through an ISO 5060:2024-controlled process, following an internationally recognised framework for translation quality assessment and applying error weighting and scaling guidelines to reflect the content’s intended use. (MQM)
Qualified linguists assess errors and their severity against agreed criteria, providing documented findings with a clear audit trail and, where commissioned, a quality score and acceptance rating.
The evaluation scope, sampling approach and scoring rules are agreed in advance to reflect the content’s intended use.
The Practical Difference Between the Three Standards
The difference is not just theoretical. It determines how the translation is produced, reviewed, controlled and evaluated.
ISO 17100
Human translation workflow
- Human translator starts from the source text
- Translation is produced without MT/AI as the production basis
- Revision by a second qualified person checks the human translation
- Best for nuance, sensitive content and full human control
ISO 18587
Full human post-editing workflow
- MT/AI output is used as the starting point
- Human post-editor corrects and improves the output
- Workflow depends on MT suitability and quality
- Best for suitable content where efficiency and control can be balanced
ISO 5060:2024
Translation quality evaluation
- Evaluates human, post-edited or unedited machine translations
- Uses agreed error categories, severity and scoring rules
- Produces documented findings and quality ratings
- Evaluation does not itself correct the text; revision or post-editing can be commissioned separately
How Termwise Applies ISO-Controlled Workflows
Termwise does not treat ISO 17100, ISO 18587 and ISO 5060:2024 as interchangeable labels. We use them to guide workflow selection, explain responsibilities and set realistic expectations before a project begins.
The right workflow depends on the content type, risk level, subject matter, intended use, confidentiality requirements, terminology complexity and the client’s quality expectations.
Human-Only Translation
For projects requiring a fully human workflow, natural target-language quality and no MT/AI use in production.
Explore Human-Only TranslationPrecision-Critical Translation & Revision
For legal, medical, technical, insurance or regulated content where subject-matter judgement and terminology control are essential.
Explore Precision-Critical Translation & RevisionAI Translation Post-Editing
Human post-editing of AI translations, with light or full review according to your requirements. Full post-editing follows an ISO 18587-controlled workflow.
Explore AI Translation Post-EditingTerminology & Termbase Management
For recurring multilingual content where consistency, reviewability and quality control need to be managed over time.
Explore Terminology & Termbase ManagementTranslation Quality Assessment — ISO 5060:2024
Independent assessment of existing translations against agreed criteria, with documented findings and, where requested, quality scoring.
Explore Language Quality AssessmentNot sure which workflow your content needs?
Send us your document, content sample or project brief. We will assess the subject matter, risk level, language combination, terminology requirements and intended use before recommending a suitable workflow.
Termwise can support ISO 17100-controlled human translation and revision, ISO 18587-controlled full post-editing, and ISO 5060:2024-controlled translation quality evaluation, alongside terminology management for multilingual content that needs professional control.