Chatting your way to a new customer experience
Chatbots are offering a new way to interact with customers. Will these improve the user’s experience or are they just another passing fad?
This article was originally published on Xwerx blog
Platforms such as Kik, Facebook Messenger, WhatsApp and Snapchat are offering businesses a new way to communicate and interact with customers via messaging apps. How can a company seize this opportunity to improve their customer’s experience?
Less apps, more messaging
Artificial intelligence and analytics have given rise to chatbots (bots for short) that enable customers to converse with an automated customer service representative via text message exchange. Bots can handle tasks such as shopping, making bookings, checking the weather and managing finances.
This move is based on trends that show users are downloading fewer apps and opting to use a handful of messaging and social networking apps on a regular basis. According toMary Meeker’s Internet Trends Reports 2016, the average user spends 80% of their time in just three of the apps they use. Meanwhile, messaging apps are “increasingly becoming second home screens”, notes the report.
Businesses need to start thinking about designing their products to be messenger first.
The West is playing catch up
Bots are the norm in Asia and the West is playing catch up. In China, 600 million WeChat users hail taxis, book appointments, and pay utility bills through the app. Recently, U.S. based company 1–800-Flowers began using Facebook’s chatbot facility. With 1–800-Flowers, customers can ask questions, get gift suggestions, place orders and receive shipping updates by having a conversation with a bot.
Uber taxi bookings, Wall Street Journal stock updates and Expedia holiday planning are also available. Once people store their credit card details within the app, users can seamlessly use bots to book flights and order food in restaurants.
The next leap in mobile user experience?
From a utilitarian perspective, bots may allow for more streamlined and faster user interaction. Take Uber for example. It allows the user to hail taxis at the tap of a button on opening the Uber app. With the Uber bot, the user doesn’t have to open the app and can order a cab by sending a short message to Uber’s bot via their messaging app of choice.
Bots also give companies a means to engage and entertain customers through appealing and personalised conversation 24/7. Bots may not always be able to solve customer requests, so real human conversation via live customer service agents should complement bot technology.
What your user needs
But, as with all new trends, remain objective and try to avoid the hype! A company’s opportunity to engage with customers via a bot must be considered beyond the sole needs of the business. Look at the context and the task at hand, while keeping in mind any other available solutions.
“Do your users really need this service to live as a chatbot?” Designer Ariel Verber had the simple idea of comparing famous chatbots with Google Search. His conclusion: “Right now, many of the bots are not better than filling a long and exhausting form”. Even if they are much cooler …
The UX design approach
Because bots are defined by pre-set scripts as well as deep learning neural networks, a bot’s responses to questions and statements can be predetermined. In addition, let’s not forget about designing a bot’s personality through careful selection of the right words, phrases and emojis.
The Lean UX method of Think, Make, Learn can be applied to bot design. By figuring out who our users are, designing the bot around their needs and validating a solution with them via testing, designers can help businesses craft successful bots.
Here are some of the UX challenges to address:
- Micro-services
The most convincing chatbots are those that address simple and straightforward transactions (e.g. order a pizza) with a few customisation options. Ask yourself if the services you provide can fit into a (quick) conversation. - Customer experience
With one more channel added to your omni-channel customer experience, you need to think at a high level. Mapping the client’s journey to identify touchpoints is critical. - Personalisation / Context
Personalisation is what makes a bot look smart. What can you learn from the user — their identity & preferences — and the context — time, location — that will reduce the friction of an automated conversation? - Anticipation
You design conversations by building flowcharts (or “scenario matrix”) that cover all potential user reactions. Right anticipation can only be achieved with the clever use of conversational patterns and deeper user testing. And because you can’t anticipate everything, always provide an escape hatch. - Conversational design
“Chatbots, by nature, need to talk like humans”, states Faisal Khalid in his provocative article“Why do chatbots suck”. How natural a bot chats may be the trickiest part to design. It means finding the appropriate tone, vocabulary and level of complexity. Once again, constant iteration — not to mention deep learning — and UX tools such as personas and user scenarios will help your bot sound human … but not too much. - Voice
A speech system is worth considering because, in many cases, it can be faster and more convenient (hands and vision free) than a graphical user interface. According to Baidu’s chief scientist, “accuracy, followed by latency” are the two key metrics to look at in this fast growing area.
The bots are rising
Bots are in the the early stages of development in the West and initial reviews have been mixed. Some bots do better than others in terms of natural conversation, abilities, response time and anticipation.
Presently, there is as much enthusiasm as there is scepticism about their usefulness, but over time, they will become smarter and more capable. It is forecast that younger millennials, aka “Generation Z”, will be the first to fully embrace bots as messaging apps are their communication channel of choice.
Time is ticking, the bots are rising, get ready for some bot banter.
By Joy Burke
