2017 UX Trends: Conversational UI
Last year we saw a paradigm shift from social networks to messaging. For the first time in history messaging apps usage surpassed social networks.
While we are currently interacting with computers via graphical interfaces which leverage mental models of the physical world represented in digital artefacts, the next wave has emerged in the form of conversational UI through messaging apps using natural language. This is done in real time where, for the first time, instead of people learning to interact with computers, computers are now learning to interact with us.
What is Conversational UI?
Conversational UI is any user interface that replicates talking with a real human. But instead of buttons, menus and drop downs you say what you want using words and emojis. In fact, it’s as simple as sending a pizza emoji to order a pizza. This can be done via voice assistants (Cortana, Google Home) or by text (WeChat), with services like Path Talk and Magic can deliver just about anything to your door, all via text message.
While industry innovators such as Messanger, Kik and Slack are already leading the charge of change in interface design, there is still much rounding of those sharp edges of technology before more organisations follow suit. However, with 50% of smartphones users now downloading zero apps per month organisations and brands will need to find a new way to connect with customers. And with the recent advancements in Artificial Intelligence, natural language technologies, and the emergence of the semantic web, 2017 could be the year to adopt.
WeChat: A Case Study
Although slow in the Western World to take off, a conversational UI messaging app in China is years ahead in this space, with the big guns in Silicon Valley scrambling to emulate their omnipresent success elsewhere. WeChat is a cross-platform instant messaging app that is comfortably sitting on top as one of the largest stand alone messaging apps in the world. First released in 2011, it now boasts an active user base of over 870 million (primarily in China).
It differs from other apps as it allows users to do a variety of tasks across multiple industries all from the one app using simple voice and text commands. It provides a whole range of features and services that can be managed and interacted with by simply saying or writing what you want. It has evolved from a text messaging, hold-to-talk voice messaging, broadcast messaging, video conferencing and sharing of photographs/videos application to a platform that has integrated social networking, banking, sharing and more.
Organisations and brands that integrate their business into Wechat are also reaping the benefits. Taxi app Didi Dache quickly doubled the number of people using its service once it integrated with WeChat, and Dior became the first luxury brand to sell top-end bags on WeChat with their first campaign selling out a limited edition “Lady Dior” bag within a day.
It’s all encompassing features allow consumers to communicate with friends, share moments and buy everything from food to paying credit card bills. And just to add to it’s impressive list users can also enjoy shopping, dating, gaming, gambling, marketing, allow users to order food, a taxi, book a doctor, find parking spaces, book hotels and flights all through it’s easy to use conversational interface and chatbots.
The Challenge for UX Designers and Organisations
So what impact does this have on design teams, development and organisations? When pushing pixels and designing graphical interfaces no longer matter, does everything change? Not necessarily. Although the basic skill set required for designing the experience may be different, some UX principles remain the same.
Creating user personas and scenarios is still a must, however, a new persona will now exist. One that represents the organisation’s brand responding to the customer. The personality of the conversational UI or chatbot becomes the UX itself (Pavlus, 2015, Johnson, NA). In some cases, this may be now the only place for brands to engage and interact with new and existing consumers so creating a likable persona and tone of voice that speaks the brands values and language is vital.
The biggest new challenge for organisations is deciding what your brand will say and do when a user prompts a question. On a human level this means understanding the context and answering appropriately and from a technical perspective it means building complex data systems that understand the structure of words, and pull from the correct data sources when needed. For engineering departments decisions will need to be made on whether to use retrieval based(Johns and Jones, 2015), descriminative or generative models (Toutanovato, 2006) to manage interactions and conversations with users.
Conversational UI interactions revolves around words so equipping design teams with the right people may become the next challenge. Companies like Howdy and X.ai have hired story tellers and dialogue experts including novelists and writers with performing arts backgrounds to improve the quality of their chatbots (Pavlus, 2016). This will help with generating more meaningful responses when interacting with users.
Brands should aim for user personalization by tracking interactions with users and systems. To do this organisations need to understand user preference well and have a mechanism to capture it. The aim should be to refine and track conversations so they can be improved over times. Upon returning to a chatbot remembering where you left off and making suggestions based on preferences and history will heighten the experience.
Summary
Although conversational UI is not a new concept 2017 may be the year where it becomes more mainstream. With tech leaders like Facebook, Microsoft and Google investing heavily in this area it’s only a matter of time before consumers and internet users (in the West) become more accustomed to using it, adapt and realise it’s benefits.
However there are still many build, measure, learn mistakes to be made before it can be accepted and trusted. However, users expectations are low so mistakes are easily forgiven so it’s a good time for organisations to experiment and trst their products in this space.
Although it comes with a new set of challenges from an organisational perspective, interaction cost and cognitive load is reduced with satisfaction and experience being improved by offering a more natural way for people to ask questions, purchase items and complete tasks .
For designers, as UI transitions from graphical to conversational the solving of problems becomes a little less tedious, demanding and confusing as it changes from design complex workflows, site maps, interactions and behaviours to focusing on conversations and meaning of context through text.
For UX Designers, the switch from designing for screens to designing for speech may be challenging but overall the end goal remains the same, we are designing for people.
“The best interface is no interface“ -Golden Krishna
Don Norman once said “The real problem with the interface is that it is an interface”. The time has come to go beyond screen based thinking. Designers have spent too long designing graphical interfaces to solve problems and users have spent too long learning new systems to complete goals. Although Don Norman’s quote and message is to use no interface, I think in conversational UI we move towards a smarter, more useful and natural way of thinking and designing, where designers spend more time solving problems and users spend less time learning new systems and more time completing goals.
References
Johnson, S. (2016, May). Chatbots and Conversational Interfaces. Retrieved January 07, 2017, from http://digintent.com/bots/
Johns, B.T. and Jones, M.N. (2015). Generating Structure From Experience: A Retrieval-Based Model of Language Processing. Canadian Journal of Experimental Psychology / Revue canadienne de psychologie expérimentale.
Pavlus, J. (2016, May) “The Next Phase of UX: Designing Chatbot Personalities” retrived from https://www.fastcodesign.com/3054934/the-next-phase-of-ux-designing-chatbot-personalities on January 5 2017.
Toutanova, K. (2006). Competitive generative models with structure learning for NLP classification tasks. Proceedings of the 2006 Conference on Empirical Methods in Natural Language ProcessingPages 576–584
