LINE Engineering
Blog

LINE's Experiences with Chatbots in Asia

Mark Serrano 2017.10.31

I'm Mark Serrano, a technical consultant for the LINE Platform.

Hi, I'm Mark Serrano, a technical consultant for the LINE Platform. On October 10th, I attended the Business of Bots conference in San Francisco as an invited speaker and panelist. In this blog, I would like to talk about my experience at the conference and also share what I talked about in my session.

The conference

The conference was held at South San Francisco Conference Center in South San Francisco, California. The attendees were from companies in many different industries from hotels, to car companies, and even sports teams. All of them came eager to learn more about chatbots and how to use them to add value to their businesses. Sessions covered an array of topics from best practices for making bots to in-depth case studies of companies that used bots to improve their businesses.

My session

In my session, I talked about some major business use cases for chatbots in LINE's major markets of Japan, Taiwan, Thailand, and Indonesia. I also outlined what I think the future looks like for chatbots in LINE's major markets.

Japan

In Japan, one popular use case that we've seen are store membership and loyalty cards. Point cards are very common in Japan and many stores, online and offline, use some sort of point card system. Most stores use a point card that can only be used at that store and as a result, customers have to have a point card for each store that they want to accumulate points for. Several businesses, including a major clothing store, have put their membership card on their LINE official accounts. Customers can just pull up the store's official account and show that to the cashier instead of having to dig through their wallet to find a card.

Customer satisfaction is huge in Japan and it's no surprise that customer support is a popular use case for chatbots. LOHACO, an e-commerce service, created a customer service bot named "Manami-san" using LINE Customer Connect to answer inquiries from their users. Manami-san first uses its AI engine to answer the questions. If Manami-san deems that the question is too difficult for the AI-engine to handle, it redirects the question to a live chat support representative to answer the question. According to this video, LOHACO reported a 90% customer satisfaction rating for "Manami-san".

One of the more well-known chatbot success stories is Yamato Transport, a major door-to-door delivery company in Japan. They created a bot to solve the "redelivery problem". In Japan, if you are not home when a package is delivered, the courier will not leave the package at your door and instead leave a slip of paper in your mailbox that prompts you to call them to reschedule the delivery. If you do not reschedule the delivery, the courier will come back every day for a week to try to to deliver your package. Yamato Transport made a bot that sends customers a notification when a package is about to be delivered and then allows them to reschedule the delivery through the bot. This benefits the customers because they can change the delivery location and time before the first delivery and reschedule missed deliveries through the LINE app, as well as Yamato Transport by increasing the chances of making a successful delivery on the first attempt. According to this video, the number of people making requests through the LINE app, such as requesting for redelivery, increased by 60%

Taiwan

In Taiwan, in addition to store membership cards, we've also seen that banking is a major use case for chatbots. Some major banks in Taiwan have created chatbots that facilitate their most common transactions like checking account balance and making bill payments.

One major bank launched a chatbot that helps users apply for home loans. It uses an AI engine to answer questions about the home loan process. For example, if a user asks "How much can I borrow?", the AI engine would ask for information like the user's age, income, and the purpose of the loan and then present them with a few loan options. The user can then start the application process through the chatbot.

Another major bank partnered with LINE Taiwan to make a credit card with LINE Pay functionality. This bank created a chatbot that provides customer service for the credit card and even allows users to apply for the credit card through the chatbot. The bot benefited users because it allowed them to apply for the LINE Pay credit card and asked questions directly through the LINE app. As a result, the bank was able to issue a large number of these credit cards. By the end of September, 2017, 750,000 credit cards were issued.

Thailand

We've seen that banking is also a big use case in Thailand. Major banks like Citibank and SCB provide services through their chatbots. Other types of bots enjoy popularity as well. Wongnai, a restaurant review service, provides a chatbot that finds nearby restaurants. Travel company, Expedia, also has a chatbot that allows users to search flights and hotels through the LINE app.

One success case in Thailand that I would like to highlight is LINE FINANCE. This is a chatbot that allows users to get information about their stock portfolios and make stock transactions. In addition to this, users are able to buy gold, which is a popular way to save money in Thailand because it serves as a hedge against the stock market and inflation. Normally, people must go to a gold shop to purchase gold where the starting price is about THB 5000 (about USD 150). LINE Thailand developed LINE FINANCE to make it easier for users to purchase gold and so they partnered with Thailand's leading gold shop to make it possible to trade gold within the LINE FINANCE bot at a minimum spend of THB 1,000 (about 30 USD). By lowering the bar for purchasing gold, LINE Thailand and the partner gold shop saw an increase in the number of transactions.

Indonesia

In Indonesia, we've seen that banking, online booking, and E-commerce are popular use cases. Banking chatbots in Indonesia are different from the ones in Thailand and Taiwan in that they usually only allow read transactions like checking account balance. Government restrictions are in place that make it more difficult for banks to allow chatbots to perform transactions that change a customer's balance.

One popular bot in Indonesia is Bang Joni, a concierge bot. Users can use Bang Joni to buy electricity tokens that are used to power their houses, buy phone call credit known as "pulsa", call an Uber, or even purchase plane or train tickets.

One of the most popular bots in Indonesia is Jemma, an AI-powered virtual friend. She engages the users in conversation and provides customized horoscopes, tips, and rewards based on what she learns about the user. The rewards that she provides usually come in the form of coupons for free goods. An example of this is that on Valentine's Day, Jemma gave out coupons for free ice cream to users that mentioned that they were in a relationship. Jemma has over 1.4 million LINE friends.

The current state of chatbots and their future in LINE's major markets

Although chatbots in LINE's major markets have enjoyed success, there is still room for improvement in terms of bot design and conversation experience. A majority of our successful use cases don't use AI or natural language processing and only respond to simple commands or simple UI buttons. Chatbots can get a good following without using AI as long as they provide functionality that is useful to their target users and as a result, developers do not believe that AI is a worthwhile investment.

Many of the bots need some improvement in areas outside of AI as well. A lot of bots do not work on creating a personality that is fun for users to talk to and many of them do not help the user return to the proper conversation flow. If users say things that the bot doesn't understand, many bots simply say "I don't understand" and end the conversation. Ultimately, the bots still feel just like bots.

Chatbots that focus completely on functionality and not at all on conversation experience are acceptable for getting early adopters, but I believe that the key to getting the users that are on the fence about chatbots is to create a more natural conversation experience. A lot of users that I talk to here in Japan believe that chatbots are just a one-sided method for companies to send information to their customers and do not even bother to converse with the bots. This belief is caused by the fact that a large number of chatbots here are used in this way. As more bot creators focus on conversation experience in addition to core functionality, and as natural language processing in Asian languages like Japanese and Indonesia improve, I believe that users will become more inclined to converse with chatbots, utilize chatbots, and understand their value.

The panel discussion

In addition to my standalone session, I was invited to participate in a panel discussion alongside representatives from Slack, Microsoft, Kik, and Google. It was an honor to share the stage with other major players in the chatbot space. We took questions from the audience that covered a range of topics like the importance of chatbots to our respective platforms, advice for creating an effective chatbot, and even each panelist's favorite chatbots. It was a great experience to listen to the kinds of questions that bot creators in the United States have, providing my perspective on them, and also hearing the viewpoints of other major players in the space.

I would like to thank the Business of Bots team for organizing such an awesome event and giving me the opportunity to speak at their conference.

LINE Platform LINE Chatbot chatbot

Mark Serrano 2017.10.31

I'm Mark Serrano, a technical consultant for the LINE Platform.

Add this entry to Hatena bookmark

Back to blog list