There is no hard limit at the moment, but it’s recommended to keep the size of each individual item under 4kB. This corresponds to 4,000 characters encoded as UTF-8 and 2,000 characters encoded as UTF-16. Uploading longer texts won’t yield any errors, but might get refused by some external translation engines integrated in Localazy.

We can gauge how good a Machine Translation Engine is by the accuracy of the translation, conciseness, consistency of output, and how many language pairs they offer.

Localazy offers four major machine translation engines to use when translating your content (available with the Additional MT feature enabled).

Use English, which is unquestionably the best language to pair with another language for machine translations because translations from and to English give the best results overall. That’s why it is also a good idea to select English as the source language for your projects in Localazy.

As for the engine itself, each one has its pros and cons:

• Google Translate is the most popular and offers many languages and language pairs with a reasonably accurate translation, making it the best average option.
• DeepL is becoming quickly better and better and is great for more technical terms. It is usually also very concise but has a lower variety of language pairs.
• Amazon Translate still has a somewhat fluctuating quality; it can be very high or pretty low, but you can use it for free with Localazy.
• Microsoft Azure has fewer available language options and average quality.

The best thing you can do is combine and compare multiple machine translation outputs and decide on the best results with backward translations and other checks.

Machine Translations (MT) are an incredible invention that makes our life easier. But sometimes, their accuracy is low, and translating large amounts of texts will inevitably produce outputs of varying quality.

Translating strings to a language you don’t know well with MT engines can be a pain because you might not fully understand what’s truly in there.

What can you do to ensure consistent quality levels of machine translation output? Localazy offers some solutions you might want to consider:

• If you are conducting the translations yourself, try to use Additional MT Engines to get more suggestions, allowing you to compare and decide.
• If you don’t want to spend your time with machine translation post-editing and checking, try the Human-Assisted Machine Translations service by Localazy and get machine translations checked for a fair price per word.

The advantages of machine translation are:

• Multiple free translation engines as options, like Google, DeepL, etc
• A myriad of language pairs, from English-Portuguese to Russian-Zulu
• Instantaneous translations. You can translate in real-time.
• Machine learning mechanisms are improving, meaning that every day the services are getting better.

The disadvantages of machine translation:

• Language pairing accuracy without English can be very low, giving it some limitations
• Risk of incorrect translation.
• Translation Engines struggle with context while translating.
• Mistakes might create irreversible situations.

How to mitigate the disadvantages of MT?

Accuracy, context, and mistakes are solvable by using Human-assisted Machine Translation and Machine Translation + Post-Editing.

You can employ both of these methods while using Localazy to achieve the best possible quality of machine translations.

When translating content, it’s common to encounter duplicate phrases or sentences. Translating these duplicates multiple times not only consumes valuable time and resources but also risks inconsistencies in translations. Here are some strategies to save on translations of duplicate content:

Use Key Clustering: Key Clustering automatically identifies and groups keys with identical values, eliminating the need to translate the same content multiple times. This not only saves time and resources but also ensures consistency across translations.

Leverage Duplicity Linking: Duplicity Linking allows you to link multiple keys with the same content. When one key is translated, all linked keys are automatically updated with the new translation. This reduces the time spent on translating duplicate content.

Utilize Machine Translations (MT): Machine Translations can provide initial translations for your content, which can then be reviewed and refined by human translators. This can speed up the translation process and reduce costs.

Remember, the key to saving on translations of duplicate content is to leverage the right tools to streamline the translation process and ensure consistency across all translations.

The quality of translations depends on how well the translator understands the source text before they begin. Localazy AI is no different from a human in this regard — both need context to increase the accuracy and quality of the resulting translation.

Localazy AI uses your glossary, style guide, key names, translation notes, comments, and previous translations to reason about what you’re translating and how it should sound. Before you translate your strings using it, prepare at least the glossary and style guide to improve results.

More providers #️⃣

There are two other machine translation providers integrated in Localazy that should take your glossary terms into account when translating:

1. DeepL integration
2. Google Translate integration

While these integrations provide helpful translation capabilities, please note that the output of these engines and their models cannot be guaranteed to be 100% reliable. We cannot take responsibility for the translations they provide. The accuracy depends on the language pair (English as source is recommended) and the provider performance.

Remember: It is always recommended to follow machine and AI translation with a human review. If you don’t have the workforce available for reviewing your MT translations in-house, you can activate our Human-Assisted Machine Translation service.

Machine Translation (DeepL, Google Translate) #️⃣

It is the fastest and cheapest option for entering new markets or testing localization. It’s useful for assessing whether a market is worth pursuing before investing heavily. The translations need significant post-editing because MT doesn’t understand context, breaks placeholders, and produces inconsistent results. But for a first iteration or localization testing, it gets content out there cheaply. Be prepared for feedback and iterate often. If you are not afraid of the potential risk of misunderstandings, you can just fix the validation issues and ship quickly by using machine translation.

Localazy AI #️⃣

Localazy AI is what you should use for initial iterations and for testing new markets when you want higher quality than basic MT. It reasons through context, preserves placeholders, applies your glossary and style guide automatically, and includes QA checks so humans can fine-tune the results, but they won’t have to fix nearly as many mistakes as with traditional machine translation. The translations come back consistent and context-aware, so your post-editing effort drops significantly. Use this for 80-90% of your content.

Professional human translators #️⃣

Heavily recommended when quality is critical. Use human translators for customer-facing marketing copy, important landing pages, established markets where brand voice matters, or content where mistakes have consequences. Humans are also still better at understanding cultural nuances and creative wordplay that AI doesn’t fully grasp.

The practical approach: Mix and match #️⃣

Start with Localazy AI for your first iteration. Context-aware translations require less post-editing than basic MT, so you can launch faster. Then decide what needs human polish. Your UI buttons and error messages might ship as-is after quick review. Your homepage headlines and key marketing content get refined by humans. You’re testing markets affordably while maintaining quality where it counts.

You can mix methods within Localazy. Set up automations that route specific languages or content types to professional translators, while the rest is processed by AI. Automation filters handle routing according to your rules.