If you’re looking to translate your website to serve customers in different languages, there’s no better technology available today than machine translation. Machine translation technology can translate text – especially large quantities of it – much faster than any human can, and at a much lower cost. It’s therefore unsurprising that the volume of machine-translated-web content has increased six times over the last two years!
While there are many machine translation systems available, two of the most reputable ones are Google Translate and DeepL. You’ve probably heard of and used Google Translate at least once in your life. But if you aren’t a professional translator, DeepL may be a more unfamiliar name.
So to help you secure the best machine translations for your website, we’re going to explore the backgrounds of Google Translate and DeepL (and their underlying technologies). We’ll also do a comparison of DeepL vs Google Translate to evaluate their pros and cons, before sharing our recommendation of the best website machine translation solution.
How did DeepL and Google Translate come into being?
To understand how DeepL and Google Translate came into being, we first need to understand the origins of their technology – namely, neural machine translation.
The very first version of machine translation was quite primitive and involved translating text according to set rules. The machine translator would pull up the relevant language pair dictionary, then use grammatical rules and direct, literal translation to translate text. However, such rules-based, word-for-word machine translation wasn’t the most accurate and often required substantial “post-editing,” or manual revision by human translators, afterward.
Next came statistical machine translation, which did somewhat better at translating text. This technology breaks the source text up into “bits” (such as words, phrases, and syntactical arrangements). After that, it crawls through massive amounts of human-translated text – known as bilingual text corpora – to find all examples of “bits” in these texts and their translated equivalents. From there, the statistical machine translation system applies prediction algorithms to churn out translations best adapted to the source text.
While statistical machine translation did a fair job of text translation, there is always room for improvement. Hence the development of neural machine translation, which uses deep learning techniques and artificial intelligence to determine the best translation for a certain text. Before being deployed, the neural machine translation software will be fed huge data sets containing different translations for a particular text, and “trained” to produce the most accurate translations possible.
At this time, neural machine translation technology is the most advanced version of machine translation available. Accordingly, it’s no surprise that you’ll find it being used in leading machine translation tools in the market – with DeepL and Google Translate being two of such tools.
So what are they, exactly?
What is DeepL? What is Google Translate?
We’ll cover Google Translate first since it’s more well-known. Operated by Google, Google Translate is probably the most popular machine translation tool used by the masses today. It translates text into over 120 languages using proprietary technology known as Google Neural Machine Translation (GNMT). While GNMT can translate text quite accurately, users are free to submit their own translation suggestions where they feel that a certain text should be translated differently.
Apart from text, Google Translate can also translate files in certain formats such as PDFs and Microsoft Word documents. It can even translate websites!
Also known as DeepL Translator or DeepL.com, DeepL prides itself on delivering “the world’s best machine translation.” Its proprietary neural networks are trained with the Linguee database to identify even the smallest nuances in text and generate the highest-quality translation possible. The business also regularly puts its translation software through blind tests to assess the quality of its translations compared to those from other providers.
Although DeepL supports fewer languages than Google Translate (28 as of writing), it has a couple of features that the latter doesn’t offer. These include:
- A translation glossary where users can create custom translation rules, and
- The adjustment of translations depending on whether you need a translation in a formal or informal tone.
What makes DeepL different from Google Translate?
Looking at Google Translate vs DeepL, there’s no doubt that Google Translate is the more popular online translator software among people not in the translation or localization industries. After all, Google Translate:
- Is very well known: If you’ve heard of Google (and who hasn’t?) then you’re likely to also be aware of Google Translate. In contrast, people who aren’t in the business of translation are less likely to have heard of DeepL.
- Supports many languages: Apart from popular languages like English, Spanish, and Italian, Google Translate can also translate text into less common languages like Telugu and Frisian (and vice versa).
- Has useful translation features like document imports, and voice, image, and website translation.
- Has a convenient integration with Google Chrome, which is one of the most popular Internet browsers today.
Accordingly, whenever a casual translator needs something translated, they are likely to make a beeline for Google Translate without considering other options.
But popularity alone doesn’t determine the best machine translation solution. So how else do Google Translate and DeepL compare?
DeepL vs Google Translate: range of support options
Google Translate and DeepL both offer a very similar range of support options.
If you need help using DeepL, you can visit its online support center. This support center contains help articles that cover everything from DeepL’s features to account management. Users of DeepL’s premium translation service, known as DeepL Pro, also have the option of submitting a support request to the DeepL team.
Similarly, Google maintains a help center for Google Translate that helps users learn how to get started with the tool and translate various media, among other issues. Users can also post their questions to a “help community” to seek assistance from community experts.
Alternatively, if you are using Google’s paid translation solution (called Google Cloud Translation) instead of the free version of the service, you can get help by filing a support case with the Google team.
DeepL vs Google Translate: supported platforms
Google Translate can be used on practically all platforms. For one, the web-based version of the tool can be used on both the Mac and Windows computer operating systems. Google Translate can also be downloaded as an Android or iOS app, making it compatible with most mobile devices and tablets available in the market.
In contrast, DeepL can be used on both Mac and Windows computers as a web-based platform or a standalone desktop app. DeepL is also available as iOS and Android apps.
Accordingly, DeepL might win slightly in terms of supported platforms for offering a standalone desktop app (whereas Google Translate does not).
DeepL vs Google Translate: translation accuracy
Translation accuracy is where DeepL starts to really stand out.
At Weglot, we recently conducted a machine translation study to evaluate the performance of various leading machine translation technologies such as Amazon Translate, DeepL, and Google Translate. We took a sample portion of text from various web pages of a company’s website, then ran this text through our chosen machine translation engines for seven language pairs. After that, we had professional linguists review the machine translations and grade them according to criteria such as accuracy and overall usability.
Our study found that of all the machine translation technologies put to the test, DeepL had the lowest number of “not acceptable” translations. In other words, most of its translations could be used without first going through a round of human revision.
Separately, DeepL conducted its own comparison study in January 2020 to see how it fared against other leading machine translation technologies. It found that DeepL performed significantly better at translating from English to German, English to French, English to Spanish, and vice versa. While Google Translate was the next best-performing machine translation technology, its ratings were nowhere near DeepL’s.
From both of these DeepL studies, it’s clear that DeepL demonstrates high translation accuracy.
How can you choose and apply the best machine translation to your business?
It’s nice to have a wide range of support options and get translations on a variety of supported platforms. But when it comes to translation, ensuring translation accuracy is probably one of your top priorities. On this basis, does it mean that DeepL is definitely the better translator to use?
Not at all.
While DeepL is undoubtedly a stellar machine translation option, there are times when Google Translate may be the better choice. For example, if you needed to translate text into a language that Google Translate supports but not DeepL, then you’d want to use Google Translate instead.
Apart from that, there may be other times when the best machine translation technology for your needs is neither DeepL nor Google Translate! In situations like these, you wouldn’t want to invest the bulk of your resources into one machine translation technology, only to find out later that it doesn’t adequately all meet your translation needs.
Instead, it makes more sense to use a translation solution that leverages various machine translation technologies to deliver the best possible translation for a particular case, no matter which technology that may be. And if it’s your website that you want to translate using machine translation, Weglot presents the perfect solution.
What is Weglot?
Weglot is a website translation solution that employs a proprietary mix of leading machine translation technologies to translate website content. These technologies include not just DeepL and Google Translate, but also Microsoft Translator and Yandex. Depending on the source text and required destination language, Weglot then picks and uses the machine translation technology that will result in the best translation.
Weglot’s machine translation engines are neural-based and are capable of “self-learning,” which means that the quality of its translations will only get better over time. We also regularly conduct internal performance tests of the machine learning technologies we use – as seen from our latest machine translation study – to ensure that we deliver to our users the most accurate and up-to-date translations. At this time, we’re able to provide high-quality translations for over 110 languages!
How can you use Weglot?
Weglot works with all leading website platforms, including WordPress, Shopify, and Webflow. As a result, you never have to worry about website compatibility issues. And once you’ve connected Weglot to your website, set up your source and destination languages. Weglot will then instantly translate all the text it can find on your website. (This includes website metadata and image alt text – our website translation solution is that thorough!)
Your translations will be stored in a central Weglot Dashboard, where you can review them and make any adjustments if you’d like to. If you manage a team or intend to engage an external language translator, you can simply add your team and/or partners to your Weglot project to grant them access to your translations.
Weglot also contains other useful features that you won’t find in standalone machine translation technologies. These include:
- Visual editor: Preview how your translations will appear on your website without leaving the Weglot Dashboard. This is handy for checking that your translations will fit within your website layout, especially since translations can differ in length from their source text.
- Media translation: Specify alternative image, PDF, and video URLs for different language versions of your web pages. Media translation is beneficial if your media files contain words, or need to be replaced to better conform to the local context.
- Multilingual search engine optimization (SEO): Weglot automatically implements hreflang tags on your website as required. These snippets of code help signal to search engines which language versions of your web pages they should be ranking higher for searchers from different countries.
- Access to a network of professional translators: If you’d like an expert to translate certain key portions of text for you, you can engage such translators from right within the Weglot Dashboard. After that, sit back and relax. The translations you’ve ordered will appear in your Weglot account when they’re ready!
Try Weglot on your website for free here.
Case studies of Weglot in action
More than 60,000 websites use Weglot to translate their content for international audiences. We can’t list all these websites in detail here, but here’s a snapshot of how some of them have benefited from making Weglot their website translation solution of choice:
Review platform REVIEWS.io helps businesses collect text and video reviews from their customers. While the platform was founded in the United Kingdom, its customer base started expanding across the globe – and REVIEWS.io wanted its Webflow website to cater to its customers’ diverse language needs.
In the beginning, REVIEWS.io tried to translate its website into German manually. However, that didn’t go well: translation accuracy was poor, and the platform found manual translation an extremely resource-intensive process. So after doing some research, the platform decided to give Weglot a try.
The results were fantastic. With Weglot, translation became an automated process that the platform could easily scale to meet their needs. German traffic to the REVIEWS.io blog also increased by 120%, which in turn led to more conversions for the platform.
Shopify ecommerce store Ron Dorff wanted a translation solution that would help it translate its store pages – including its entire catalog of over 150 products – with ease. And the store didn’t take too long to realize that Weglot was the way to go.
That was because once the store had installed the Weglot Translate app via the Shopify App Store, it had a quality first pass of translation of its website content in just a few minutes. The store subsequently took fewer than 15 days to review and refine Weglot’s translations for use. Implementation of the finalized translations was also a breeze, as Weglot’s no-code solution meant that Ron Dorff didn’t have to get a developer to push the translations to its live website.
And while translating your website can lead to poorer search engine rankings if mismanaged, Ron Dorff faced no such issues. Weglot took care of the technical task of helping search engines identify the appropriate language versions of Ron Dorff’s pages to show to searchers.
The year that Ron Dorff started using Weglot, it experienced a more than 400% increase in website visits to its ecommerce store, plus more than 70% growth in international sales.
DeepL vs Google Translate: what’s the verdict?
We’ve seen how DeepL and Google Translate are both powerful machine translation tools in their own right, each having their own pros and cons.
For one, Google Translate is highly accessible and supports a wider variety of languages. At the same time, DeepL has shown itself to be the more accurate machine translation service, with it also being able to provide more nuanced translations. Either way, both Google Translate and DeepL are backed by the most sophisticated machine translation technology available today, namely neural machine translation.
If you’re torn between DeepL vs Google Translate for your business’ website translation needs, why settle for only one of them when you can have access to both – and other leading machine learning translations? Our Weglot website translation solution incorporates Google Translate and DeepL, plus Microsoft Translator and Yandex, to provide your website with the highest-quality machine translations.
After completing a first pass of machine translation with Weglot, you can optionally go on to add a human touch to your translations before approving them for use. And if you need specialized translation expertise for this purpose, Weglot makes it easy to invite your preferred translation agency to your translation project or engage a partner translator through the Weglot Dashboard.
As seen from the case studies above, Weglot is seamlessly compatible with popular website platforms, and has helped businesses in various industries increase their traffic and conversions. Your business can experience these benefits, too. Simply sign up for a free Weglot trial to get started.