Machine Translation Post-Editing (MTPE) is review and editing of machine-generated target text against a defined brief. It is not simply proofreading because the source, output and intended use must be considered together.
Suitability varies with source quality, language pair, subject matter, audience and risk. Some material is better handled through human translation from the source.
Who is this service for?
The service may be relevant to the following project owners when the material and requirements fit the defined scope.
Teams with existing machine-translated content requiring assessment
Content operations managing high-volume, lower-risk material
Product or support teams working with repeatable text sets
Organisations comparing MTPE with human translation for a defined use
Detailed use cases
These scenarios show how audience, inputs, scope factors and limitations interact; they are not promises that every project is suitable.
Knowledge-base output
A support team has machine output for repetitive help articles. Inputs include source, raw output, glossary and audience. Scope depends on output quality and formatting. MTPE may not suit safety-critical or highly ambiguous content.
Internal content batch
An operations team needs usable internal material from machine output. Inputs include purpose, tolerance and representative samples. Scope changes with error density and source quality. The agreed post-editing level limits the expected finish.
Specialised content assessment
A team considers MTPE for technical or regulated material. Inputs include terminology, risk owner and sample output. Scope depends on subject complexity and traceability. Human translation may be preferable where machine errors create disproportionate risk.
What shapes scope, timing and pricing?
A quotation is based on the reviewed material or session requirements, not the service name alone.
Timing, availability and pricing are confirmed only after the relevant inputs and dependencies have been reviewed.
Source quality, clarity, consistency and file preparation
Raw machine output quality and the system-independent error profile
Intended audience, use, content risk and post-editing level
Terminology, format, volume, sampling and requested date
How this service differs
MTPE starts from machine output and follows an agreed editing brief; human translation starts directly from the source. The efficient choice depends on actual material, not volume alone.
Light and fuller post-editing have different intervention goals. The required level should be defined in practical terms for the audience and use.
How to prepare
These steps help expose dependencies before quotation.
Provide the source and unedited machine output as separate files
State audience, intended use, risk and acceptable intervention level
Supply terminology, style references and protected content
Share a representative sample before assuming the whole set is suitable
Limitations and responsibilities
Poor, ambiguous or inconsistent source text can reduce MTPE suitability and increase review effort.
MTPE cannot remove all risk from machine output; suitability and scope are confirmed only after material review.
Information needed for a quote
Share the details below when you request a quote so the scope can be reviewed properly.
Source language
Target language
Project date
Project details
Estimated scope
Optional file attachment
How requesting this service usually works
Submit paired source and outputProvide matching source text and raw machine translation without pre-mixing undocumented edits.
Define use and editing levelState audience, content risk, terminology and what the edited text must be fit to do.
Assess a representative sampleReview error patterns, source quality and format to judge suitability and likely effort.
Choose and quote the workflowConfirm MTPE level or recommend direct human translation where that is more appropriate.
Edit against the agreed briefWork follows the accepted baseline; changes in source, output or required finish trigger reassessment.
MTPE Suitability and Risk Triage
Support a reasoned choice between MTPE levels and human translation.
Purpose: Support a reasoned choice between MTPE levels and human translation.
Who it helps: Content teams holding machine output and workflow owners managing risk.
How to use it: Score a representative sample for source clarity, error density, terminology and consequence of error.
Limitation: It is a scoping aid; quality remains dependent on the material, brief, domain and risk review.
Step 1
Source: ambiguity, consistency, formatting and terminology
Step 2
Output: omissions, mistranslations, fluency issues and token errors
Step 3
Decision: audience, consequence of error, editing level and alternative workflow
Clear, consistent source text and usable machine output may support MTPE, particularly where audience, risk and editing expectations are defined. Suitability is assessed from the material.
How does source quality affect MTPE?
Ambiguity, errors and inconsistent terminology in the source can produce unstable output and increase editing effort.
What post-editing levels are available?
The required intervention should be defined in the quotation brief, for example a limited fit-for-purpose edit or a fuller edit. Labels alone are insufficient without acceptance criteria.
When is human translation preferable?
It may be preferable for highly nuanced, creative, sensitive, safety-critical or poor-source content, or where sample output shows that editing is not an efficient or suitable route.
Should raw machine output be supplied?
Yes. Provide aligned source and unedited output, plus any prior edits separately, so the starting point can be assessed.
Can MTPE outcomes be guaranteed?
No. Scope and suitability depend on source, output, language pair, subject matter, audience and risk.
Ready to discuss your project?
Tell us about the languages, materials and intended use. Requirements are reviewed before the scope is confirmed, and pricing is based on the confirmed scope of work.