Google Translate debuted in 2006 as a multilingual neural machine translation service, engineered to translate text, documents, and websites from one language into another. Today, more than 500 million people around the globe use the service to communicate and decipher foreign languages for everyday conversations.
Businesses also turn to Google Translate for assistance in providing customer support or to translate their websites and digital marketing collateral. But if your business values the customer experience, then generic machine translation alone is not enough.
Is Google Translate accurate?
The first thing companies want to know when evaluating Google’s translation service for business use: “Is Google Translate accurate?” As a general-purpose solution, Google Translate is simple to use and helps bridge the gap between foreign speakers. It is able to, more or less, help facilitate communication and understanding. (Emphasis on more or less.)
The problem with Google Translate is that the quality of the translations isn’t always accurate. Anyone who has used it can attest to the textbook, oftentimes, word-for-word translations that fail to truly capture the essence behind the words. And forget about idioms.
There are countless idioms that exist in one country and not the other. Have you ever tried using Google to translate the phrase let’s play it by ear from English into Spanish? We did, and unless you are planning to (physically) play with your customer’s hearing, such literal translations become meaningless.
In fact, even within a country, people living in different regions or on opposite coasts use words in different contexts. Let’s say you are a native Japanese speaker trying to translate a message for an English-speaking audience; it can be challenging to deliver 100% accuracy without knowing exactly where the English speaker resides, and the variations of the language in that area. An attorney in the United States is referred to as a solicitor in England.
French speakers in France may use the same words as those in Quebec — same as Portuguese in Brazil vs Portugal — but words and phrases could have vastly different meanings or even no meaning at all. Google Translate simply does not support businesses’ Lokalisering needs.
Context is key; personalization — paramount
Beyond geographic localization, words can be “local” to a business or industry. Take the word clutch, for example. A fashion company would use it to describe a type of purse, while an automobile company would use it when referring to a car part. It’s an issue where context is paramount, and Google is solely trained on broad domain data, relying on the infinite amount of Internet data it indexes in its search engine to answer a wide range of translation queries.
Ooni Pizza Ovens ran into a problem of contextualization when attempting to translate words like pizza pie, pepperoni, pizza peel, and even wood-fired pizza flavor into Italian. The translations (more like mistranslations) did not land properly, and even caused some offense in the case of wood-fired pizza flavor, where some customers believed such a thing would lead to cancer. A glossary unique to Ooni’s business would have helped. (And it did!)
It behooves businesses not to rely on Google Translate, as such inaccuracies can cause irreparable damage to their brands. Customers expect and even demand native-language communication when spending their money with a company — we found that 57% of consumers consider it a bias when brands don’t offer end-to-end multilingual experiences to their customers.
A little thing called data privacy
Users may be unaware that they relinquish a number of data privacy rights when using Google Translate on their browser. That’s because Google can harvest information from translated texts and use it to improve its own algorithms. What does this mean for businesses who trust Google Translate to translate sensitive contact, payment, or healthcare data?
Well, it could mean breaching internal data privacy protocols and government regulations such as HIPAA. Team members may not understand how relying on Google Translate can make their organization susceptible to data privacy violations and associated fines. The snippet below from Google reveals just which rights users grant Google when leveraging their product.
So what else is out there?
While Google Translate is an easy, DIY solution that is free to use and can be accessed via a web page, app, or Google API, it wasn’t engineered to integrate specifically into your business’s CRM. This means that in the absence of a custom build, your support team members might have to click back and forth between applications, cutting and pasting messages, and hoping for the best that the translations are good enough to wrap up the customer inquiry. This is slow, inefficient, nowhere near scalable, and no way to run a successful business.
So what are your options? A few alternative popular translation technologies include:
DeepL: A machine translation-only, general-purpose solution that is not designed specifically for customer support and is useful for FAQs, but not emails or chat.
Language I/O: A predominantly machine translation solution that includes human review only as a paid add-on.
Lokalise: A predominantly machine translation solution billed monthly or annually and per character. Human review available as a paid add-on.
Unbabel: AI-powered, human-refined translation purpose-built specifically for customer support. Unbabel supports FAQs, emails, and chat, all with flexible pricing.
MT designed specifically for customer’s brands
While they are all built on the foundation of machine translation, Unbabel is unique in that it features an AI that is enhanced with human editors and machine learning so that the machine translation is constantly improving to deliver more accurate translations over time.
Language is more of an art than a science. With that, Unbabel leans on its global community of human editors to localize translations, finessing words and phrases so that translated messages are aligned with the local culture. They work diligently in the background to ensure your written customer support interactions sound less like a textbook and more like the way people in the specific local markets you serve actually talk. From Arabic to Vietnamese, to Turkish to Korean, they currently support 29 languages with the ability to tailor each of them to the local landscape.
Unbabel easily integrates with popular CRMs including Salesforce Service Cloud, Zendesk, Intercom, and Freshdesk. The technology embeds seamlessly into your customer service and marketing workflows so team members can start using it with minimal disruption and training. In fact, most customers on the receiving end of a translated message will never know that a (human-assisted) machine sits between them and a customer support agent. Both parties can simply chat in their native languages, and Unbabel’s AI outputs the right words for flawless communications.
And although the integrations are out-of-the-box, their translations are anything but. Organizations can customize the translator to their specific business or vertical by adding glossaries and style guides into their Unbabel portal. For example, a gaming company may want to include words for specific in-game characters, weapons, currencies, or other objects that are unique to their game, so that Unbabel is fully prepared and can anticipate member queries and responses.
Plus, it was built with data protection at the forefront of mind — it is fully GDPR and CCPA compliant, with robust data encryption and anonymization to ensure the privacy of both businesses and their customers.
Schedule a demo to experience the technology in action, and see why companies all around the world trust their translations to Unbabel.
See Unbabel in action:
