Why translation pricing is changing

TL;DR:

Advances in technology and shifting business expectations are making traditional, word count–based translation pricing increasingly outdated. Now, price-per-word models fail to account for differences in content complexity and value, especially with the rise of AI-generated and machine-translated texts. As a result, language service providers are adopting hybrid pricing strategies, including per-character, per-line, and hourly or project-based rates, to better reflect the true cost and quality required for different translation tasks.

Translation pricing for human translation, machine translation, and post-editing usually follows a simple formula: count the words, apply a rate, and send the invoice. It was predictable, easy to compare, and easy to budget.

This model is under pressure. Expertise, technology, risk, and changing expectations have made price per word too simplistic to reflect the real work involved.

On the client side, the expectation remains the same: a transparent and predictable pricing model, which word pricing allows. On the side of language service providers (LSPs), however, switching to hourly rates for jobs like post-editing is a question of fairness. Post-editing time varies significantly based on the quality of the machine output, and a flat per-word rate doesn’t account for that variability. As a result, the current pricing model is evolving.

Why price per word no longer reflects translation value

The traditional model assumes all words carry equal weight. A legal contract and a product description are treated the same if they contain the same number of words. That has never reflected reality.

What’s changing now isn’t the existence of price per word, but its dominance. AI-generated content and machine translation have exposed the gap between volume and value. If machines can produce text instantly, the real cost is no longer in generating words. It’s in ensuring those words are correct, appropriate, and safe to publish.

This shift is why many language service providers (LSPs) now use hybrid pricing:

Per-word pricing for straightforward content in editable formats
Per-character pricing for Asian languages where word boundaries aren’t clearly defined
Per-line pricing, which remains the standard in European markets
Hourly or project-based pricing for services that can’t be measured by text volume
Fixed programme fees for ongoing multilingual operations

At SwissGlobal, the standard unit for human translation and machine translation with post-editing is per line, not per word. One standard line consists of 55 characters, including spaces. This cost method is the established convention in Switzerland and much of Europe, where a standard page contains 30 lines, or between 1,500 and 1,800 characters. Words are used only when a client specifically requests them, typically for documents in editable formats like Word, Excel, or PowerPoint, where word counts are precise and easily extracted. Find out specifically how much a translation costs in one of our previous blogs.

Per-character pricing isn’t a niche exception. For Chinese, Japanese, and Korean, it’s the standard model because these languages don’t use spaces to separate words. A single character in Chinese can carry the meaning of an entire English word, making word counts arbitrary and misleading. LSPs typically charge per source character for these languages, with rates varying based on complexity and target language. If your business operates in Asian markets, understanding this distinction is essential for accurate budgeting and vendor comparison.

Price per word still exists because it is easy to understand and easy to approve internally. Procurement teams, in particular, aren’t eager to give up something that fits neatly into a spreadsheet. In 2026, per-word rates for common language pairs still range from $0.10 to $0.22, and for many businesses, this remains the default comparison metric during vendor selection.

AI may lower some costs, but it adds others

There is a persistent expectation that AI will dramatically reduce the cost of translation services. It can. Sometimes.

Machine translation with post-editing, for example, can reduce costs for high-volume, lower-risk content. In 2026, MTPE services typically cost 40–60% less than full human translation, since translation productivity typically doubles or triples: post-editors can handle around 5,600 words per day compared to 1,600 for traditional translation with quality assurance.

At SwissGlobal, we offer machine translation with expert revision to balance efficiency and quality, certified under ISO 18587 for machine translation post-editing.

However, focusing only on cost reduction misses the bigger picture. AI introduces new cost factors:

More complex quality control
Data security requirements
Ongoing terminology management
Rework when AI output misses the context
Time invested into prompt optimisation

Poorly implemented AI workflows often shift costs downstream. What looks cheaper upfront can become expensive when content needs correction after publication.

Translation is no longer a one-off task

Translation used to be a discrete task. A document went in, and a translated document came out. That model no longer reflects how businesses operate.

Content now lives across multiple systems: websites, apps, support platforms, and marketing tools. It is constantly updated, reused, and repurposed. As a result, translation pricing is increasingly tied to systems and infrastructure, not just individual projects.

Automation and AI can reduce manual effort but require setup, maintenance, and oversight. The key question is no longer whether to use AI, but how to embed it intelligently into existing workflows. The goal is an end-to-end process that removes inefficiencies, such as moving content between disconnected systems just to apply AI and push it to the next stage.

Multi-service workflows are now standard

Translation rarely happens in isolation. A typical project includes:

Project management: coordination of timelines, resources, communication, and delivery across all stakeholders
Translation and/or machine translation with post-editing
Revision: a second linguist reviews the translation for accuracy and consistency Translation quality assurance: systematic checks for terminology, formatting, omissions, and technical consistency
File preparation and post-processing: preparing source files for translation tools and restoring them into usable formats after translation
Desktop publishing
Terminology management

At SwissGlobal, these are defined services with specific scopes:

Machine translation with post-editing combines automated translation with expert human revision. The machine generates a first draft; a qualified linguist then corrects errors, improves fluency, and ensures terminology accuracy. This balances speed and cost with quality, and is certified under ISO 18587 for machine translation post-editing
Translation by a qualified native-speaking translator in accordance with ISO 17100 applies the six-eyes principle, which adds a dedicated revision step. A second linguist reviews the translation in depth, adding an extra layer of quality assurance before delivery.
Revision is a comprehensive quality check beyond proofreading, including accuracy, consistency, terminology, and alignment with the source text.
Proofreading is a monolingual review of spelling, grammar, and punctuation only. Linguistic editing includes proofreading plus improvements to style, clarity, and flow.

Each layer adds value. Each layer adds cost. Evaluating this using a single metric like cost per word is no longer sufficient.

Compliance and governance now affect translation pricing

AI becomes more embedded in translation workflows, and governance is no longer optional. Businesses need to define acceptable quality levels, how AI tools are used and controlled, how data is handled, and who is accountable.

These quality levels are particularly relevant in Switzerland and across Europe, where GDPR directly influences how translation services are delivered. SwissGlobal maintains infrastructure certified to ISO 27001 for information security.

Regulatory pressure isn’t limited to Europe. In the United States, new ADA requirements mandate that public-sector digital content meet WCAG 2.1 Level AA accessibility standards. This content includes translated web content, documents, and multimedia. For businesses serving public institutions, accessibility compliance is becoming a direct cost factor.

Certified translations, notarisation, and apostille services remain largely unaffected by automation. These are fixed, regulated processes where pricing reflects legal requirements. At SwissGlobal, these are offered as package services that include notarisation, apostille where required, and shipping.

The procurement reality most predictions ignore

Procurement teams still prefer transparent pricing. Even if more sophisticated models better reflect reality, businesses often default to price per word for comparability and fixed rates for budgeting predictability. This approach creates tension: LSPs are building complex, value-driven pricing structures while buyers still compare everything using spreadsheets designed for a different era.

The result is a fragmented market. Premium LSPs offer structured, multi-layered services with human oversight and certified quality management. Lower-cost vendors rely heavily on automation with minimal oversight, sometimes at rates that undercut standard market pricing significantly. Understanding where your business sits between these options matters more than chasing the lowest rate.

What businesses should expect next

Translation pricing won’t change overnight. However, the direction is clear: more hybrid pricing models, clearer segmentation of content by risk level, greater transparency around workflows, and a stronger link between price and business impact rather than volume alone.

A more realistic way to think about translation pricing

Translation is no longer just a production cost. It’s part of how your business communicates globally. That doesn’t mean every project needs a complex pricing model. In many cases, simple pricing still works. However, relying on outdated assumptions can lead to poor decisions: overpaying for low-risk content or underinvesting in high-risk material.

At SwissGlobal, the focus is on aligning the service with the purpose. That may involve efficient machine translation with post-editing, or full human translation with multiple layers of quality control. It may also involve addressing content quality before translation begins through services like plain language editing or style guide development. We help businesses choose the right mix of human translation, MTPE, and quality assurance.