Optimizing Translation Budgets with Machine Translation (with Examples)

Optimizing Translation Budgets with Machine Translation (with Examples)

This post was published on Sep 8 2022 and updated on Mar 5 2024 to reflect the latest trends in machine translation.

If you read our last blog post, you’ll know all about the trials and tribulations of using unedited machine translation outputs. In this article, we’ll show you how to optimize your translation budget with machine translation, without sacrificing quality. 

We spoke to three clients about how they use machine translation to boost cost efficiency while still producing high-quality content. The three companies were: 

  • Snowflake: a pioneering Enterprise Data Management software provider who paired machine translation post-editing with workflow automation to achieve 36% cost savings. 
  • A global company specialized in talent management software that combined neural machine translation with human quality assurance for cost-effective translation of large volumes of content. 
  • A world leader in contact center solutions who defined standards for their source text to produce better machine translation outputs. 


Related article: Snowflake’s Technical Manual Translation Blueprint for Cost Reduction 

Challenges with content localization in a digital era 

In the era of big data, many companies struggle to efficiently manage massive volumes of content. This problem is exacerbated when this content needs to be localized on a lean translation budget.

Despite being from different industries, all three of our clients shared the same challenge: the need to provide international users with online product documentation in their native languages. 

In dealing with this high demand to localize large volumes of content, their key objectives were to ensure:  

  • Translation costs did not skyrocket
  • The quality of translated content did not deteriorate
  • Translations are delivered punctually

This is precisely where machine translation enters the picture. 


What is machine translation suitable for? 

All three companies strategically utilized machine translation for product documentation, help content, and software UI. This aligns with the industry trend of relying on machines for swift translation of large content volumes, which would be cost-prohibitive with human translations.

Importantly, this content is typically technical, prioritizing comprehensibility over stylistic perfection. For instance, in texts like technical instructions, accuracy is crucial, and literary finesse is not necessary.


Common issues with machine translation 

Machine translation promises incredible cost savings of up to 50 percent. But as we’ve said before, publishing raw outputs that have not been revised by a human expert can backfire. This is something all our clients raised as a concern.

Common issues cited about raw machine translation

  • Contextual understanding: While neural machine translation has propelled output quality, it is often unable to capture the underlying meaning and context of information. 
  • Readability: Machine translation doesn’t always render the correct word order needed to make a text read fluently. For example, if three compound nouns are used in one sentence, machine outputs can be incredibly clunky, thus impacting user-friendliness. 
  • Formatting issues: Document types with special formatting elements like XML and PowerPoint don’t play well with machine translation systems. This often results in word-for-word translations that are almost identical to the original text, misplaced line breaks, or other formatting complications. 
  • Coherence & cohesion: Machines don’t possess the same text comprehension abilities as humans. This means languages with grammatical constructions like gender are not handled particularly well. For example, while objects are simply referred to as “it” in English, this “it” could be “sie”, “er”, or “es” in German. 

Well-written source text is crucial for accurate machine interpretation and the production of high-quality translations. Our contact center client improved machine translation quality by adhering to clearly defined writing standards. Adjusting their writing style enhanced localization efforts by reducing post-editing time and thus maximizing their translation budget.

To improve the grammar, readability, and fluency of raw machine translations, another method that emerged over the last year is to run outputs through LLMs (large language models), such as ChatGPT.  


Translation tip: When selecting a machine translation provider, look for an ISO 18587:2018 certification for guaranteed accordance to international machine translation post-editing standards. 


Stretch your translation budget with flexible quality levels

When working with machine translation, a best practice to manage your translation budget is to define the quality levels required. For example, internal documents may require less stringent quality assurance than an external user help document.

Our client from the talent management industry had large amounts of online help documentation to localize. Budget constraints ruled out traditional workflows and conventional post-editing was also too expensive for their massive content volume.

Localization managers found themselves caught between upper management’s cost-saving demands and international customers needing localized software instructions. Simultaneously, the sales team wanted high-quality materials to help them close deals.

Although raw machine translation excelled in cost efficiency, publishing content without human revisions proved too risky.

After thorough consultation and extensive testing with Milengo, the client chose a solution involving a neural translation system and a human quality assurance step. The solution focused on ensuring correct software UI and using predefined key terminology. Overall, it enabled significant translation cost reduction while maintaining essential quality criteria.

Milengo collaborates with clients to identify key areas in projects where accuracy and fluency are crucial. Based on this, we curate a machine translation process to achieve the required quality within their budget.

Find out more about our AI-powered machine translation services 


Balancing cost, quality, and time 

To demonstrate the cost-saving potential of machine translation post-editing, Milengo and Snowflake embarked on a pilot project. This allowed both parties to identify areas of importance, rectify potential challenges, and ensure the solution was scalable

To make things trickier, Snowflake’s weekly release cycles meant automation and streamlined processes were essential to delivering content on time. To tackle this, we paired a rigorously tested machine translation system with highly automated translation management processes to speed up workflows and reduce manual file handling. 

Results from this pilot were used to develop a specialized workflow that adhered to their business needs. The outcome: a tailored machine translation process that delivered the highest quality translations in the shortest possible time, cutting translation costs by around 36%.  

Get all the details: Download the Snowflake case study


Boost translation ROI with Milengo’s machine translation services 

Businesses are turning to machine translation due to growing volumes of content, short delivery timelines, and high cost pressure. Additionally, technological developments enhance its appeal, making it a go-to solution for specific content types.

Our clients achieved reduced translation costs and improved quality by implementing varying levels of human quality assurance on machine translation. This revision step depends on content type, as printed technical device instructions demand higher stylistic requirements than troubleshooting articles in a knowledge base.

Regardless of the approach, it’s evident that machine-supported translation solutions striking the perfect balance between quality and cost are the future of localization.

At Milengo, we don’t offer off-the-shelf machine translation solutions. We take a consultative approach to ensure clients have the best machine translation setup for their company’s localization needs within their budget. Interested in maximizing your translation budget and boosting ROI? Schedule a call with our localization experts.

Smarter translation with AI

Sophia Guan

read all posts

After 11 years in 9-5 roles, Sophia ditched routine to embrace a digital nomad life. With a diverse background in global communications, a strong cultural curiosity, and a passion for bridging cultures and people, she fosters meaningful connections through effective, creative, and unique cross-cultural communication.