October 14, 2020

From your website to the world: training a bot in multiple languages

More and more companies are using conversational bots to help their customers. In fact, Global Market Insights estimates that the chatbot market will exceed more than 1.3 billion dollars by 2024.

One thing's for sure, AI-powered bots are excellent tools for responding to frequently-asked questions, making personalized suggestions, and helping customers 24/7 across multiple channels.

Nevertheless, there's one essential characteristic that no customer service bot can be without: the ability to chat! 

That’s why speaking your users’ language is the first step in creating everyday and fluid conversations. Here we’ll explain how you can create an intelligent bot that speaks multiple languages at the same time. 

Speak in your customers’ language

Imagine you’re in charge of customer service at a company called The Awesome Store that sells products online throughout different English-speaking countries.

Obviously, they all speak the same language. But they'll use different words to say the same thing. An American might greet you with a "hi dude, how's it going?”, while a British person would say “hey mate, you alright?”. 

In this case, what you first need is for your bot to speak English and understand its regionalisms so it doesn’t end up giving “oops” as an answer. 

But things get trickier if you also have customers from Mexico, who speak an entirely different language. 

The solution? Create polyglot conversational bots that can be set up in multiple languages and, thanks to artificial intelligence, can also understand natural language

Conversational AI in charge

Offering an efficient experience means a lot of things, like 24/7 responses, omnichannel communication and personalized suggestions.

But if you have a chatbot, there's one more issue to handle: chatting with your customers naturally like a human. 

What does this mean? Understanding typos and spelling errors, emojis, regionalisms, and obviously speaking the same language as your customer. 

There are all kinds of chatbots, but not all of them can do this. Only the ones developed with conversational artificial intelligence will be able to engage in natural conversations that truly add value. These can also detect the text’s true intention and understand queries no matter how they're asked.

Related article: Conversational engine vs. conversation flows: What's the difference?

So...how do you create a bot that can chat in multiple languages, using regionalisms and adapting to each customer’s language?

Training a polyglot virtual assistant

Let's look at some techniques you can implement in your customer service strategy in order to train your multi-language virtual assistant.

1. Know your customers

Without knowing your customers, there's no strategy. And without strategy, there’s not a good experience.

That’s why the first step is knowing who the users are on your website, what their main needs are, and in this case, knowing which countries or regions they come from. This will point you in the right direction to know which language your content should be in, both within and outside the bot.

2. Create content tailored to your users

Once you know your users, you can go on to creating content. First, you have to make sure the bot responds to all frequently-asked questions. What is it they’re looking for when they chat with it? Business hours, shipping status or product stock? Include all relevant information!

Remember that your virtual assistant’s success depends upon giving it all the tools it needs to understand your customers. That’s why it’s super important to consider all the ways a user could ask a question, keeping regionalisms in mind. 

So within the same question you can include all the words you want your bot to understand.

AgentBot, Aivo’s conversational chatbot, has a customized dictionary for each company’s needs.

Let’s look at an example: if you offer a credit card for premium customers, like Trust Bank Gold, you can crear a union of words. In this way, each time someone mention that combination of words, the virtual assistant will understand it and it will offer the right answer.

Also, you can manually add specific regionalisms from each country. For example, let’s pretend you have a clothing store. You sell to both customers in the USA and the UK. For the word “jeans”, you can also add “pants” for american users, and “trousers” for english users. 

3. Implement one bot per language

Let’s go back to The Awesome Store example. In the previous point we just solved the challenge of being able to chat with customers from different English-speaking countries.

Now there’s another challenge for your virtual assistant: being able to communicate with your Spanish-speaking customers from Mexico. Don’t worry, there's also simple solution for this.

First of all, the bot’s engine needs to be able to adapt to different languages. AgentBot, for example, can be configured for up to 157 languages, including Spanish, English, Portuguese, French and Italian.

Once you have this, half your work is already done.

The second step is requesting your new bot in the language you need, and placing it on the same language website. That means a company will have as many bots as languages it needs, since each language’s bot is set up and hosted separately.

Amanda, the virtual assistant of Amadeus, speaks Spanish, English, and Portuguese, and she’s currently learning Chinese and German. The same thing happens with Nimbi’s chatbot, which speaks three languages fluently. 

Related article: Nimbi automates 90% of queries in three different languages

4. Offer a unified experience

Once you have your bots in different languages and with all the regionalisms included, there’s one final step: unifying the experience.

No doubt you already have your website set up so each customer automatically accesses the content in their own language.

But they might accidentally go to the wrong website and find the bot speaks to them in another language. Our advice? Always be one step ahead and solve these problems beforehand for a better experience.

A good practice for this is leaving prepared answers in case a user speaks a different language. In your answer, you can include a link to your website in the language they request. Here's an example from Aivo’s virtual assistant, Aibot:

Are you ready to have an international bot?

In an increasingly global market, being able to automate customer queries from multiple regions is crucial for improving your presence.

At the same time, speaking the same language as your customers is indispensable for engaging in more human and long-lasting relationships.

If you want to know more about how our solutions can help you achieve this, don’t wait to schedule a call with one of our representatives. 

Are you looking for new ways to improve your CX?

Our customer service solutions powered by conversational AI can help you deliver an efficient, 24/7 experience  to your customers. Get in touch with one of our specialists to further discuss how they can help your business.