Data shows that, for global businesses, providing support in multiple languages is well worth the effort. Nearly three quarters of people search online in their native language, which means that if you’re only communicating in English, for example, you’re probably losing customers and adding layers of inefficiencies for your agents.
Easier said than done, perhaps. On average, even in one language, 20% of agent time is spent looking for information to either share directly with customers or to find the right way to resolve a problem. Providing support in multiple languages across multiple channels adds another set of variables to the mix. Offering multilingual support on live channels especially can get costly and inefficient, because you’ll need to hire a global, multilingual staff, implement a real-time translation solution, or both.
With customers all over the world opting to self-serve, turning your knowledge base into an international resource can have a big impact on your customers and agents. Having one source of truth for knowledge is a scalable, time-saving solution that fosters a more accurate and consistent experience for your multilingual customers.
A multilingual help center has been a central element of success and a key component of the global support strategy at FINALCAD, for example. The FINALCAD app is available in 30 languages and the company offers help center content in 10 of those languages. A regional approach was also essential for OLX, a classifieds platform that does business in more than 40 countries and employs more than 1,200 people worldwide. Given the company’s global footprint, the team was able to reduce the number of tickets submitted by creating market-specific self-service articles within their knowledge base. Globally, this led to a 40-percent reduction in ticket volume.
Another thing to consider is the difference between translation and localisation. Localised content goes beyond translation, perceiving nuances in patterns of speech, intent, concepts, or phrasing. Localised content knows to call it a spanner in the U.K. and a wrench in the U.S. It knows where they call it soccer and where they call it football. And it can identify different uses of bank—the financial institution vs. the side of a river.
This is where AI can come into play. A lot of language nuances are tough to aggregate and localise at scale, which is why a language-trained model is designed to absorb some of those headaches. Language-trained AI models get smarter over time, learning and recalibrating every time it makes a mistake. The idea is not only equipping agents with a global resource, such as a knowledge base, but using that as a basis for leveraging technology that can learn to perform its job even better over time.
Language models first try to understand the words, context, and intention behind the question, and then tries to match those variables with the most relevant article speaking to that problem. Zendesk’s Answer Bot, for example, is available in English, Spanish, Portuguese, German, French and Dutch. Each language has a different model underpinning its functionality, negating the need to translate the content each time. It gets the difference between soccer or football because it drills down into how a word is used in context by both customers and agents. It then gets to work finding and suggesting relevant articles from your help center.
Time to value is one of the main challenges in implementing any new software solution, especially for growing and time-strapped departments. One of the benefits of a language-based AI model, is that it works immediately, out of the box, understanding the intent behind customer requests and matching those requests to content in your help center.
If you’re an international business, multilingual customer support is essential. Hiring agents in every region isn’t always a practical or scalable solution, which is why a multilingual knowledge base can support your agents provide consistent support, wherever they are assisting customers.