By now, DeepL needs no introduction. When I first heard about it, around five years ago, DeepL was touted as an even better version of Google Translate.
And it hasn’t failed to live up to its name.
In 2018, TechCrunch praised DeepL for being more accurate and nuanced than other machine translation tools in the field, claiming it had outdone other tech giants (quite an achievement when you’re up against giants like Google) and raised the bar for machine translation.
Today, DeepL claims to be the “world’s best” AI translation, and has the results to prove it.
But with new AI translation tools changing the machine translation landscape, is DeepL really still the world’s best translation tool? Or are the machine translation tides about to change?
Keep reading to find out.
What is DeepL?
Ok, you already know it’s a machine translation tool.
But to go even more granular, DeepL is a neural machine translation (NMT) tool that uses neural networks to mimic the human brain.
Launched in 2017 in Germany, DeepL was the brainchild of Jarek Kutylowski, a computer scientist originally from Poland. Jarek moved to Germany at the age of 12 without speaking a single word of German. This was the moment he realized how important language was, which later led him to launch DeepL.
What does the ‘L’ in DeepL stand for?
Good question. It stands for ‘learning’.
Deep learning is a subset of artificial intelligence (AI). It aims to simulate the human brain so that outputs are more accurate. For translation that means higher quality and more natural-sounding machine translations.
How does Deepl translate?
As mentioned earlier, DeepL is a neural machine translation tool.
This means that it considers entire sentences, not just individual words, to grasp the full context of content and deliver more accurate translations.
To illustrate the importance of context, consider how politicians manipulate quotes to distort meanings.
During the 2012 U.S. presidential campaign, President Barack Obama said, ‘If you’ve got a business, you didn’t build that.’ This remark was widely misconstrued by opponents, who portrayed it as an attack on entrepreneurs.
By cherry-picking words, you warp intended meanings.
A tactic that mirrors the limitations of older machine translation methods. These tools rely on limited word patterns and fail to capture the complete context of phrases or sentences, resulting in inaccurate translations.
In contrast, neural networks learn from extensive datasets and analyze entire sentences to understand the broader context, which is crucial for effective translation and localization. By processing full sentences, they deliver more accurate translations.
What additional translation features does DeepL have?
DeepL has a bunch of handy features that help you translate faster, more accurately, and on-brand.
Let’s go into each feature in more detail.
Language support across 30 languages
DeepL has come far in its 7 years of existence. It started with translations between 7 European languages, and now supports 30 languages with over 800 possible combinations between them. Including Russian, Portuguese, Chinese, Japanese, and most recently Arabic.
API access
DeepL API gives businesses direct access to its machine translation technology. This means they can connect it to their websites or applications to deliver a seamless translation experience.
Data security for DeepL Pro customers
Your data will not be stored and used if you use DeepL Pro to translate texts. They will only be kept temporarily for the production and transmission of the translation. After which they’ll be deleted. Texts will not be used to improve the quality of DeepL’s services unless you use the free version:
We process your texts, the documents you upload, and their translations for a limited period of time to train and improve our neural networks and translation algorithms. This also applies to corrections you make to our translation suggestions.
Glossary
There’s the option to add a glossary, allowing you to set rules for how words and phrases are translated. There might be short phrases that appear often in your text, or brand terms or names you don’t want to translate. It’s easy to add one or more terms to your glossary in the interface, with the term in your source language (source text) and your preferred translation (target text).
Top quality for various industries
DeepL seems to perform well across the board. It has a lot of enterprise customers spanning from B2C to B2B across various countries and sectors. The glossary feature has helped improve the quality for users who can now define custom terminology.
Informal/ formal tone
When it comes to translations for different brands, situations, or regions, you might need to adapt your tone. DeepL offers you the choice between translations in a formal or informal tone of voice. When you’re translating languages where pronouns differ for informal and formal, these features can come in very handy.
For example, in informal situations in French, the pronoun tu is preferred instead of vous.
Document translator
DeepL has an easy-to-use, real-time online document translator. You can translate your PDFs, Word Docs, PPTs, and more with a single click while preserving their original format.
Free plan available
With a free plan, a familiar interface, seamless integrations with Windows and iOS, and a Chrome plugin, DeepL is accessible to everyone. However, the number of glossaries and terminology entries is limited on the free plan.
AI writing companion
While not a translation feature, DeepL has recently launched an AI writing tool, DeepL Write. It can help you find variations for phrases. A use case in translation might be that you ask for variations of your source language, then plug them into DeepL’s translation tools.
Is Deepl better than Google Translate?
DeepL and Google Translate are the most popular machine translation (MT) engines on the market.
While Google Translate keeps getting smarter, and they both run on an NMT model, DeepL is still considered better quality than Google. You also may be wondering – is Google Translate accurate? How does it compare to Deepl?
DeepL is often preferred because it goes beyond word-for-word translations, and captures more context and nuances of the original text, making translations feel more natural and fluent than their GT’s output.
The upside of Google is that it supports way more languages (133!) and, like DeepL, is free to use. But, the glossary feature, which is one of the most important features for translators and linguists, only appears to be available through the API.
Both DeepL and Google Translate have very friendly user interfaces, where you can translate text or documents, and translate up to 500,000 characters per month for free.
One really cool feature that Google offers is its website translation feature.
Just pop your URL into Google Translate and choose the language you want to translate from and to:
Open up the link to your website and you’ll see the translated version:
So which is better? It really comes down to language support. If your languages are supported by DeepL, then DeepL is generally considered to be better for its more accurate output.
Who are DeepL’s other competitors?
The rise of generative artificial intelligence (AI) tools, like ChatGPT, has led to a boom in translation technology, and Statista predicts that the market size for AI translation will grow from about $6bn in 2024 to about $12bn by 2030.
So soon, DeepL will start to see a lot more competition from newer AI translation tools. Here are some old and new DeepL competitors to watch out for in 2024:
Lokalise AI
Lokalise AI is a generative AI translation tool that helps teams translate large volumes of content quickly and accurately. It comes attached to a TMS (translation management system), so you can better manage all your translation projects and connect to more modern tools, like content management systems, design plugins, repositories, and more.
You can add more than a glossary to Lokalise AI to improve translation accuracy. Upload your style guide, add further context, and even use translation memory to store and reuse already approved translations. Additionally, Lokalise is your go-to TMS for both translation and localization.
Microsoft Translator
Bing offers text and speech translation via a simple user interface that’s similar to DeepL and supports just over 100 languages.
To translate text into more than one language, you’ll have to copy and paste your text and translate it one by one.
Microsoft Translator offers a few other cool features like automatic language detection, widely used phrases in different languages, and the option to listen to the translation and download the app.
Bard
Another one from Google, with a ChatGPT-feel and completely free, for now.
Just add your prompt, asking Bard to translate content into your target languages, and your results will appear instantly.
Bard even preserves formatting and you can upload an image and ask Bard to write a description and translate it.
Copy.ai
With its Chat-GPT-style user interface, Copy.ai is easy to use. Used primarily as an AI writing companion, it also has impressive translation capabilities. Simply plug in your prompt, asking it to translate content into your target languages. You’ll need to specify the countries for different dialects, for example, Brazilian Portuguese or Chilean Spanish.
You can add your brand voice guide to make sure translations are on brand, then hit ‘enter’ and you’ll instantly receive translations.
Is DeepL worth it?
Again, it depends on your use case. The consensus is that DeepL is one of the best machine translation tools but you’ll need to test it for your use case and langage pair.
Let’s look at some clear advantages and disadvantages of using machine translation tools like DeepL to help you understand if DeepL is worth it.
Advantages of using DeepL
Speeds up the translation process
DeepL helps language professionals save huge amounts of time, so they’re better positioned to meet tight deadlines. According to Konstantin Dranch, Founder at CustomMT, a translator can boost output from roughly 2,000 to 4,000–8,000 words per day while maintaining a good level of quality, if they switch to post-editing of machine-translated texts.
Easy to use and quick translations
The great thing about DeepL is that it’s easy enough for everyone to use. There’s no learning curve, just paste in your translations and you’ll get translations back in seconds.
Not too expensive
Using machine translation instead of relying fully on human translators will save you time and money.
DeepL offers a free version with limited features, allowing users to translate texts of up to 500,000 characters a month. However, the free version may not be suitable for commercial or high-volume translation.
Several paid options go up to 57,49 $US a month. They all offer unlimited text translations with the price changing based on the number of glossary entries and files you can upload.
There’s also the API Pro plan, which gives you unlimited character limits, and the possibility to integrate with other tools
Good enough for some content
Not all content is created equal. You might not always need the highest quality translations.
Content like internal documentation, customer support articles, and FAQs don’t need perfect style and consistency. For most content of low visibility and importance, DeepL will likely be sufficient to get the job done.
For straightforward content, like translating a list of country names, you could happily rely on DeepL alone.
Integrates with other tools
DeepL can be integrated into larger translation workflows managed by other platforms or systems. That way you don’t need to copy and paste translations from DeepL to your website or apps.
Disadvantages of using DeepL
Placeholders not preserved
Placeholders are a set of symbols, replaced by dynamic content like names, addresses, honorifics, birthdays, or directions:
“Hello %s”
Sometimes MT engines like DeepL fail to recognize placeholders. One solution is to split your text and translate it separately, but you might lose valuable context that Deepl needs to translate your content accurately.
If placeholders are corrupted during translation, in the best-case scenario your text sounds awkward or is difficult to read. In the worst-case scenario, your app will crash. There are many different placeholders, each with its own syntax (i18n, ICU, printf, and so on), so you need to set up a process that catches these before production.
No translation management ecosystem
DeepL doesn’t offer translation management capabilities like those found in dedicated translation management systems (TMS). While DeepL offers an API that developers can use to integrate its translation services into their applications or workflows, it does not come with valuable translation management features (translation memory, option to add a style guide, upload visual context, etc) that streamline translation workflows.
Lacks creativity for marketing content
DeepL lacks the creativity to nail marketing localization on the head. Point in case:
Translations often sound robotic and don’t always match the original tone of the content.
For high visibility/high priority content like website copy, marketing or advertising copy, and sales collateral, you’ll want to be sure that your company’s distinct style and voice come across just right.
No translation memory support
DeepL doesn’t have a Translation memory (TM). A database of sentences, or segments of text, and their translations that can be automatically reused when translating similar or identical content.
Everything that you (or any other team members) type in the editor, upload, or set via an API is saved automatically for future use. It can save businesses a lot of time and money when included in the translation process.
Whether DeepL is worth it depends on several factors, highlighted below. The key is to balance your budget, time, technical capabilities, and quality standards.
- The type of content you’re translating: Are you primarily translating simple text such as buttons or other product UI needs? Is content low-risk? If yes, then DeepL is a cost-effective solution that can work well when combined with human translators.
- The translation quality you need: Does language quality meet your company’s standards? Is your content low visibility or high visibility? The more visible your content, the better the quality you’ll need. You could still use DeepL but with a human in the loop.
- How quickly do you need content: Do you need to launch an offer tomorrow or do you have more time? DeepL can translate a lot faster than human translators, but can you compromise quality? These are the questions you’ll need to ask before choosing a translation approach.
- The volume of content you need to translate: can you manage translations yourself or do you need help? Huge volumes of content are better managed with a translation management system alongside DeepL. You don’t want to rely on DeepL alone for huge volumes of content, otherwise, you’ll spend hours copying and pasting translations from one tool to another (among many other inefficient workflows that lead to headaches).
- Your budget: Can you afford human translators to do everything or not? Huge cost savings in terms of time and money might be a good enough reason for you to choose machine translation post-editing (MTPE) over human translation.
The best way to manage translations with DeepL
One way we recommend using DeepL is alongside translation management systems (TMS) that offer more robust management capabilities. These tools typically allow you to organize, track, and manage translations, collaborate with translators, maintain translation memories and glossaries, and streamline the translation workflow.
Most TMS tools make it easy to integrate DeepL into your translation workflow using the API.
Here’s what your DeepL + TMS workflow could look like:
In an ideal world, your content will live in a content management system (CMS) and connect to your TMS through integrations or an API.
While it’s fairly common for companies to have multiple systems for managing content, often a TMS is the only place where all your multilingual content gets stored – in all your target languages, along with the original content.
Lokalise is a TMS that integrates machine translation services, like DeepL, to translate your content. You can also connect DeepL through your API.
Lokalise also connects to many other tools, like content management systems, code repositories, design plugins, and more, so you can save yourself the headache of flipping between tools just to copy information from one tool to another.
You can use templates to automate your tasks, move even faster, and reduce the manual work. Add to that, a glossary, translation memory, and the ability to preserve placeholders, and you’ve got yourself a very powerful ally in your translation process.
There’s also the option to use an AI translator inside Lokalise TMS if you need more creative translations than what you get with DeepL and Google Translate. With Lokalise AI, you can upload your style guide, add context, and use it alongside your glossary and translation memory, giving you translations that are on-brand and accurate.
You can also ask for alternatives and shorten translations with just one click, saving you time trying to come up with options that fit.
There’s even an AI ‘editor’ inside Lokalise, which you can use to review your translations. Why not use it to determine the quality of your DeepL translations?!
To translate with DeepL or not?
The simple answer is, yes. DeepL is an excellent machine translation tool. However, it also depends on your language pair – DeepL doesn’t support all languages. You should also compare the outputs of other translation engines because some perform better when dealing with certain content types.
The best approach is to play around with different translation engines and see which one works best for your language pair and content type. That might mean you end up using them all.
While DeepL is considered one of the best machine translation tools on the market, you should keep a very close eye on AI translation tools, which are powered by large language models (LLMs) like OpenAI.
As AI continues to evolve at breakneck speed, you can expect to see businesses invest more in AI translation tools. These tools won’t just be powered by one translation engine either. They’ll be powered by several LLM models to deliver the most accurate translation for that content type and/or language pair.
Want to get ahead in the world of AI translation? Try Lokalise AI for free.
