Machine translation accuracy 

How accurate is machine translation for website translation?
Workflow automation illustration

Trusted by thousands of leading global brands

united nations
IBM
tupperware
Mary Kay
doordash
hbo
nielsen

MT and website translation

Let’s face it, you likely don’t have time to look into the accuracy of machine translation and even if you did, the studies aren’t always evaluating based on the content you’re translating.

That’s why together with industry-leading Nimdzi Insights we did that study for you. By using human linguists, we looked into the performance and accuracy of the top 5 machine translation providers and exactly how reliable they are for website translation.

Read on for the results.

“Machine translation has reached a level of maturity that makes it possible to translate high-volume content that otherwise would not be translated because of budget and time constraints – complementing the human linguist.”

Roman Civin

VP of Consulting at Nimdzi Insights

“We take advantage of neural machine translation in our 37 languages and our global team makes quick fixes if needed. It changed our way of working and we love it!”

Eric Espinoza
Product Manager

IBM logo

“Immediate translations accelerated our conversions and let our merchandisers focus on other tasks. We now have more flexibility and launch available stock much faster.”

Matthieu Galtie
Head of Digital Marketing

“Machine translation gave us an instantly multilingual website. We were able to modify the translations and then publish the content faster than with a traditional translation method.”

Albert Baqués
Marketing Manager

Weglot users save as much as 80% of time spent on website translation thanks to machine translation.

“Machine translation can be deployed as an accelerator in the website translation process.”

The study concluded that 85% of the machine-translated content sample was very good or acceptable in terms of usability.

“NMT as a technology has reached a certain level of maturity”

Out of the 168 translation segments sampled, the following chart identifies per language how many did not require editing and were used as they were.

“Contemporary NMT is considered highly usable for website translation and requires only minor editing.”

At no point did the linguists have to make edits to more than 30% of machine translations across any of the language pairs and NMT providers.

Machine translation conclusions

Expand

MT is fit for translating marketing content like company websites.

Challenge

To overcome perceptions, managers need to understand true MT capabilities. 

Optimize

There’s no “winner MT”, there are advanced choices for best business impact.

Abandon

A localization model that relies on post-editing as a process step is old news.

Hear from companies using machine translation

READY TO GET STARTED?
Try Weglot for 10 days and see how you really can have an instantly translated site in minutes.
background shape
background shape