Machine translation has come a long way. What once produced clunky, barely readable output now generates drafts that are often 70–90% accurate — accurate enough that a skilled human translator can refine them into polished, publication-ready text in a fraction of the time it would take to translate from scratch. That process is called Post-Edited Machine Translation (PEMT), and it has become one of the most cost-effective strategies in the modern translation industry.

But as businesses and procurement teams start exploring PEMT, one question comes up almost immediately: How does pricing work? More specifically, should you pay per word or per hour? The answer is not as straightforward as most providers make it sound, and choosing the wrong model can leave you either overpaying or underwhelming your translators — both of which hurt the final quality of your content.

This article breaks down how each PEMT pricing model works, where each one excels, and what factors you should weigh before committing to a structure for your next project.

What Is Post-Edited Machine Translation (PEMT)?

Post-edited machine translation is a hybrid workflow that combines the speed of automated machine translation (MT) engines — such as DeepL, Google Translate, or Microsoft Translator — with the precision and cultural fluency of a human editor. A machine translation engine processes the source text first, generating a raw translated draft. A professional linguist then reviews that draft, correcting errors in grammar, terminology, style, register, and cultural appropriateness before the content is delivered to the client.

PEMT is not the same as simply running a document through a free translation tool and calling it done. It is a structured, quality-controlled process. Depending on the content type and its intended use, post-editing can be performed at two levels: light post-editing, which targets factual accuracy and basic fluency, and full post-editing, which produces output that meets the same standard as human translation. Legal contracts, regulatory filings, and marketing copy almost always require full post-editing, while internal communications or low-stakes informational content may be suitable for a lighter touch.

Because post-editors are working with an existing draft rather than producing a translation from a blank page, they can typically handle a higher volume of words per hour compared to traditional translation. This efficiency gain is where PEMT’s cost advantage comes from — and it is also why pricing these projects requires a different framework than standard translation jobs.

The Two Main PEMT Pricing Models Explained

Per-Word Pricing

Per-word pricing is the most familiar model in the translation industry. With PEMT, the client pays a set rate for every word in the source or target document. PEMT per-word rates are typically lower than standard human translation rates — often ranging from $0.03 to $0.08 per word for full post-editing, compared to $0.10 to $0.20 per word for human translation, depending on the language pair and subject matter. Light post-editing rates can be even lower, sometimes falling between $0.01 and $0.04 per word.

The appeal of this model is transparency. You receive a word count from your source file, multiply it by the agreed rate, and you know your budget before the project begins. It is easy to compare quotes across providers, easy to explain to finance teams, and easy to scale — if your document grows by 500 words, the cost adjustment is straightforward. For clients managing large volumes of content across website translation or multilingual documentation, this predictability is extremely valuable.

However, per-word pricing has a structural weakness in the context of PEMT: it does not account for the actual condition of the machine translation output. If the MT engine produces near-perfect text in a well-supported language pair like English to Spanish, the post-editor’s job is minimal, yet they are paid the same rate as if the output were deeply flawed. Conversely, if the MT output for a technical or rare language pair is poor quality, the post-editor may spend significantly more time on each word than the rate compensates for.

Hourly Pricing

Hourly pricing charges the client based on the number of hours a post-editor spends working on the content, regardless of word count. Rates for professional post-editors typically fall between $25 and $75 per hour, depending on the language, subject matter expertise, and level of post-editing required. This model has historically been more common in software localization, transcription services, and desktop publishing workflows where the billable work cannot be neatly reduced to a word count.

The primary advantage of hourly pricing is that it accurately reflects the effort involved. If the machine translation output is poor — riddled with mistranslations, awkward phrasing, or culturally inappropriate expressions — the post-editor is compensated fairly for the additional time spent. This model also tends to attract more experienced linguists who are confident in their productivity and do not want to absorb the risk of slow MT output through a fixed per-word rate.

The downside for clients is unpredictability. Without a clearly defined word count ceiling or a project cap, hourly billing can balloon beyond the initial estimate, particularly for complex content types. Some clients are also uncomfortable approving an open-ended scope, especially for large or ongoing projects. Strong project management and regular check-ins become essential when working on an hourly basis.

Hourly vs Per Word: A Side-by-Side Comparison

Choosing between the two models often comes down to a few key dimensions. Here is how they compare across the factors that matter most to translation buyers:

• Budget predictability: Per-word pricing wins clearly here. You know your maximum cost before the project starts. Hourly billing requires trust in the provider’s time-tracking and a well-scoped brief.
• Fairness to linguists: Hourly pricing is more equitable when MT quality is inconsistent. Per-word pricing can undercompensate editors on difficult language pairs and overcompensate on easy ones.
• Scalability: Per-word scales effortlessly for high-volume content such as e-commerce product catalogues or multilingual knowledge bases. Hourly is better suited to smaller, more complex projects.
• Content complexity: Highly technical or creative content — such as pharmaceutical documentation or localisation projects requiring cultural adaptation — is often better priced hourly, as MT quality for these content types is more variable.
• MT engine quality: If you are using a well-trained, domain-specific MT engine with a strong track record for your language pair, per-word pricing becomes more defensible. If the MT engine is general-purpose or the language pair is underserved, hourly pricing protects both parties.

Factors That Affect PEMT Pricing

Regardless of which model you choose, several variables will influence the actual rate you are quoted. Understanding these factors helps you have a more informed conversation with your language service provider and set realistic budget expectations from the outset.

Language pair: High-resource language pairs with well-trained MT engines (English to French, Spanish, German) will generally produce cleaner output and attract lower per-word rates. Low-resource or typologically distant pairs (English to Thai, Japanese, or Arabic) often require more intensive editing and command higher rates under either pricing model.

Subject matter and terminology: Legal, financial, and pharmaceutical content involves specialized terminology that general MT engines frequently mistranslate. Post-editors working in these domains need subject matter expertise, which is reflected in their rates. If your business operates in these sectors, factor in this premium — and expect that a rigorous quality review, similar to the grammar proofreading and editing process offered through professional proofreading services, may be needed after post-editing as well.

MT engine quality and domain training: A translation memory-integrated, domain-trained MT engine will produce substantially better output than a general-purpose free tool. Better MT output means less post-editing effort, which translates to lower costs under either model.

Level of post-editing required: Light post-editing is faster and cheaper. Full post-editing, which brings the text to human translation quality, takes longer and costs more. Be honest with yourself about which level your content actually requires — over-specifying will waste budget, while under-specifying can damage your brand or, in regulated industries, create compliance risks.

File format and layout complexity: Content embedded in complex InDesign files, HTML templates, or structured XML requires additional handling. Desktop publishing and typesetting work often runs as a separate line item but can affect the total project cost significantly.

When to Use Each Pricing Model

A useful rule of thumb is to match the pricing model to the nature of the content and the reliability of the MT output. Per-word pricing works best when you have a high volume of relatively straightforward content, a well-supported language pair, and a consistent MT engine producing predictably decent output. Think: large-scale e-commerce catalogues, user manuals, or internal knowledge base articles translated into major European languages.

Hourly pricing is the safer choice when the content is complex, creative, or highly technical; when the target language pair is less well-served by major MT engines; or when the scope of post-editing work is genuinely uncertain before the project begins. Marketing copy, legal agreements requiring precise terminology, or content for culturally nuanced markets in Southeast Asia are good examples where an hourly model may serve both client and linguist better.

Some providers offer a hybrid model: a per-word rate with a minimum hourly guarantee or a cap. This can be an effective compromise, particularly for ongoing relationships where both parties are building a track record of MT output quality and editor productivity together.

Hidden Costs to Watch Out For

PEMT pricing conversations often focus on the post-editing rate itself, but the total cost of a PEMT project can include several additional items that are easy to overlook at the quoting stage.

• MT engine setup and API costs: If your provider is running a commercial MT engine or deploying a custom-trained model, there may be setup fees or per-character API charges that are passed on to the client, either explicitly or embedded in the per-word rate.
• Translation memory (TM) leverage discounts: Reputable providers apply TM matches to reduce costs on repeated or similar content. Make sure your quote reflects these discounts and ask how matches are tiered (fuzzy vs. exact match).
• Quality assurance and review steps: Full post-editing ideally includes at least one independent review pass. If this is not included in your quoted rate, budget for it separately.
• Project management fees: For complex, multi-language projects, project management time is often billed separately, particularly under hourly models.
• Revision rounds: Clarify upfront how many revision rounds are included. Unlimited revisions are rarely sustainable for the provider and are usually not what the contract actually means.

Choosing the Right PEMT Provider

The pricing model matters, but it is only one dimension of a good PEMT partnership. The quality of the post-editors themselves — their domain knowledge, language proficiency, and familiarity with your industry’s terminology and regulatory requirements — will ultimately determine whether PEMT delivers the return on investment it promises.

Look for a provider with a structured quality assurance process that includes not just post-editing but also independent editing and cultural review. In markets like Singapore and the broader Asia Pacific region, cultural appropriateness is not a minor consideration — it can determine whether a product launch resonates or falls flat. A provider with demonstrated expertise across a wide network of certified linguists, covering both high-resource and regional language pairs, is better positioned to deliver consistent quality across diverse projects.

It is also worth asking potential providers about their MT engine strategy. Do they use a general-purpose tool, or do they work with domain-trained engines suited to your industry? How do they handle language pairs where MT quality is likely to be lower? What does their escalation process look like when post-edited output does not meet the required standard? These conversations will tell you a great deal about how seriously a provider takes quality — and how transparent they are willing to be about cost.

For businesses requiring professional language translation services across multiple content types and markets, partnering with a provider that can offer both PEMT and full human translation — with a clear, honest recommendation about which approach suits each project — is the most reliable path to quality outcomes and controlled costs.

Making the Right Call on PEMT Pricing

There is no universally correct answer to the hourly vs per-word debate in PEMT pricing. Both models have genuine strengths, and the right choice depends on your content type, language pairs, MT engine quality, and how much budget certainty you need going into a project. Per-word pricing offers transparency and scalability; hourly pricing offers fairness and flexibility when the scope is less defined.

What matters most is working with a provider who is transparent about how their pricing is structured, what is and is not included, and how they ensure the post-edited output meets the quality standard your content requires. Ask the right questions upfront, understand what level of post-editing your content actually needs, and do not let a low per-word rate distract you from the total cost and quality picture. When PEMT is done well, it is one of the most powerful tools available for scaling multilingual content without sacrificing accuracy or brand integrity.