Plan Ahead to Build Successful Multilingual Chatbots
Chatbots are rapidly becoming ubiquitous in marketing and support. The potential for brands to interact with customers using natural language – and perhaps a bit of personality – without needing an army of paid human agents is driving major investment from enterprises. Tech giants – from Google to Facebook and IBM and Weibo to Microsoft – have started a virtual race to dominate this field. However, what is missing in this picture so far is serious attention to multilingual needs. Amazon, Apple, Google, and Microsoft have all released digital assistants – Alexa, Siri, Google Assistant, and Cortana respectively – in multiple languages, but these are the exception, and other enterprises struggle to move past the language of their home market.
As part of a detailed examination of how enterprises provide multilingual customer support, CSA Research interviewed market leaders in the creation of chatbots, also known as intelligent online agents, about how they provide multilingual capabilities. The results revealed that few companies have found ways to deal with language in a systematic fashion. As a result, results are sporadic, methods are ad hoc, and failures common. One chatbot implementer at a large tech vendor – one that develops a commonly used open chatbot framework – stated that the company’s team expects a 30% success rate from its own efforts.
Chatbots pose challenges fundamentally different from what is seen with traditional content. The shift to conversational structures and the need to embrace “messy” terminology are among these.
As you build – or localize – chatbots, CSA Research’s examination of the issues reveals several techniques you can adopt to help ensure success. Five key points are:
- Incorporate linguists into the development phase. When you have language experts involved with development, they can help you spot your assumptions and plan for how to handle them. Enterprises should reach out to localization partners early on for advice. LSPs who are aware of pending chatbot projects should reach out and ask to provide feedback.
- Do not rely on machine translation. MT is an attractive option that can work well for simple bots, but as they grow in complexity, it cannot solve the cultural issues that your chatbots are likely to face. If you do choose to use it, terminology management becomes especially critical. But for all but the simplest of bots, you would still need to adapt the scripts and frameworks for each locale.
- Set realistic expectations. You can successfully localize skills bots that follow defined scripts, but more complex conversational agents are orders of magnitude more complex and may rely on language technologies that are not available for many languages. Be prepared to scale back or adjust your plans as needed in response to the technologies and chat platforms available in the various markets you support. LSPs should learn about the chat applications in common use in their language markets so they can advise clients about them and their strengths and weaknesses.
- Develop specifications early on and stick to them. Chatbot projects are often prone to feature creep because of their open-ended nature. For each bot you develop, define exactly what it will – and will not – do and then follow these guidelines. Do not add features late on: Each new one you include increases the complexity and the likelihood that the finished product will not work in your target markets. LSPs should help their clients understand why shifting the goal posts late on is likely to result in major problems and expense.
- Plan ahead for security, legal, and regulatory issues. Different markets have different expectations and laws concerning these issues. Either design your bots to comply with the strictest requirements you will face in any market, or plan to adapt your localized versions for each market. If you do not do so, you run the risk of legal penalties. Although LSPs should not dispense legal advice, they can advise their clients about what issues they need to consider and can help connect them to proper legal counsel if required.
About the Author